diff --git a/newLang/docs/architecture.md b/newLang/docs/architecture.md
index 53fcdcd..b56064a 100644
--- a/newLang/docs/architecture.md
+++ b/newLang/docs/architecture.md
@@ -710,10 +710,11 @@ past the end value) and `b = 3`, with no recorded print output. The
 method that increments it again, and explicitly returns; the 53 emitted bytes
 leave `n = 2`.
 
-Word values, call arguments, arrays, method result forwarding, the frame
-prologue, and sizing/pass-3 addresses remain unimplemented. `for` currently
-requires a constant byte end and step 1 and leaves the loop variable one past
-the end value. E04-S07 owns those limits and the 255-byte code buffer.
+Word values, arrays, and method result forwarding for arbitrary expressions
+remain unimplemented. `for` currently requires a constant byte end and step 1
+and leaves the loop variable one past the end value. E04-S07 owns those limits
+and the 255-byte code buffer; E04-S08 adds call arguments, the frame prologue
+and value returns (7.13).
 
 ### 7.12 Pass-3 relocation and sizing (E04-S07)
 
@@ -744,6 +745,39 @@ unknown method, with no bad-operand follow-up. The class fixture checks
 `METHOD DESTROY` and `METHOD STUB` MAP records, and the output fixture
 reconfirms the E202 overwrite refusal.
 
+### 7.13 Calling-convention emission (E04-S08)
+
+`emit.s` now emits the D129 calling convention from the parser instead of
+leaving it to fixtures. `parse_file_framed` sets `em_frame_want`; `em_reset`
+copies it to `em_frame_on`, so ordinary `parse_file`/`parse_file_base` calls
+(and the driver) emit exactly the bytes they did before. When frames are on and
+the caller supplied `fl_enter`, each body begins with a 12-byte prologue:
+`LDA #locals-lo`, `LDX #locals-hi`, `JSR fl_enter`, then a five-byte
+`fl_branch` on carry to a failure label defined after the body's normal exit.
+`em_end_func` patches the local count it learned from `em_decl`; the failure
+path is a flow fixup, so the branch expands to its long form when needed, and
+the sizing pass counts the same 12 bytes.
+
+`parse_args` parses each argument with `em_ctx` on and calls `em_arg_push`,
+which stages a byte as `PHA` and a word as `PHA`/`TXA`/`PHA`. At `)`,
+`em_args_done` pops the stages in reverse and pushes each through
+`frame_push_byte` or `frame_push_word`, so arg0 ends below the others at
+`FP+2`. `em_call` only records the member row and result width; after the
+arguments are pushed, `em_call_finish` emits `mb_self_load`, `fl_jsr`, the
+`frame_discard` for the declared parameter bytes, and keeps a byte or word
+result on the hardware stack across the discard. `ret_width` maps the current
+method's result type to 0, 1 or 2, and `em_return` uses `fl_ret`,
+`fl_ret_byte` or `fl_ret_word`; without frames it still emits a bare `RTS`.
+
+`args.n8` declares `Calc.twice`, which returns `self.scale(3)`, and
+`Calc.scale(n as byte)`, which returns `n + 3`. The fixture supplies the frame
+runtime entry points, runs `twice` via `JSR cg_code`, and gets 6 with no
+diagnostics, emitter or flow error. The `calls.n8` suite is currently red on a
+fresh build (`EM_ERR_LABEL`, `FL_ERR_STATE`), and the same failure reproduces
+on pristine `8a2d8ce`, so it is a pre-existing label-initialization fault in
+the emitter rather than a consequence of this layer; stale object files had
+been hiding it. Fixing it is separate work.
+
 ## 8. Heap and Ownership
 
 - The generated-program heap initially resides in the same RAM bank as program
diff --git a/newLang/docs/decisions.md b/newLang/docs/decisions.md
index a67bb39..dcaaa79 100644
--- a/newLang/docs/decisions.md
+++ b/newLang/docs/decisions.md
@@ -40,6 +40,7 @@ documented default. `Proposed` means a later change still needs approval.
 | D116 | Runtime faults report a stable code and the best available map-resolvable program address. | Rich stack traces are deferred; faults remain diagnosable within the runtime budget. |
 | D117 | Freeze the compiler memory map in `cfg/compiler.cfg`: 8K bottom common RAM, COMMON at `$1300`, private zero page at bank 0 page `$1A` via P0, bank-1 tables at `$2000-$BFFF`, IRQs masked and NMIs guarded in compiler states, test mailbox at `$1BF0`. | Internal contract backed by E01-S05 x128 evidence (`docs/memory-map.md`); no user-visible behavior except that RESTORE is ignored while compiling. Generated-program maps remain open. |
 | D118 | The compiler loads the whole source file once into a bank-1 cache (`$2000-$6AFF`) and replays it for every pass, instead of re-reading the disk. | Rewind is exact by construction and avoids reading a slow disk three times. Compiler tables get the remaining 21,760 bytes of bank 1 (`$6B00-$BFFF`). Internal; no user-visible behavior. |
+| D156 | Emit the calling convention from the parser (E04-S08): arguments are evaluated left to right and staged on the hardware stack, then popped in reverse through `frame_push_byte`/`frame_push_word` so arg0 lands at `FP+2`; a 12-byte prologue (`LDA`/`LDX` locals, `JSR frame_enter`, `BCS fail`) is reserved at each body when frames are requested and its local count is patched after the body; `return` goes through `frame_leave` with a 0/1/2-byte result preserved. Frames are opt-in through `parse_file_framed`, so existing fixtures that do not supply frame entry points emit unchanged code. | The generated program now uses the D129 frame ABI for real calls instead of only in hand-written runtime fixtures. The prologue reserves fixed bytes and patches the count, because locals are known only after the body is parsed; the sizing pass sees the same 12 bytes, so label offsets and lengths still agree. A refused frame branches past the body's normal exit and returns without `frame_leave`. Word arguments and byte/word results are covered; interface arguments and four-byte returns still belong to E06-S07. |
 
 ## Proposed
 
diff --git a/newLang/planning/claims/E04-S08.md b/newLang/planning/claims/E04-S08.md
new file mode 100644
index 0000000..a0ac790
--- /dev/null
+++ b/newLang/planning/claims/E04-S08.md
@@ -0,0 +1,123 @@
+---
+story: E04-S08
+owner: codebuff
+status: done
+started: 2026-11-10
+completed: 2026-11-10
+depends-on: [E04-S07]
+---
+
+# E04-S08 Claim
+
+## Intended change
+
+Emit the D129 calling convention from the parser: stage each call argument on
+the hardware stack, pop the stages in reverse through `frame_push_byte` /
+`frame_push_word` so arg0 lands at `FP+2`, reserve a patched frame prologue at
+the start of each body, and return through `frame_leave` with a 0-, 1- or
+2-byte result preserved. Frames stay opt-in so existing fixtures that do not
+supply frame entry points keep their previous code shape (D156).
+
+## File area
+
+- `src/common/emit.inc`: `EM_ERR_ARG`, `EM_ARG_MAX`, `EM_PRO_SIZE`
+- `src/compiler/emit.s`: `em_prologue_start`, `em_arg_push`, `em_args_done`,
+  `em_call_finish`, `em_arg_total`, patched prologue and result-width return
+- `src/compiler/flow.s`: `fl_push` / `fl_push_word` cells
+- `src/compiler/parser.s`: `parse_file_framed`, argument staging hooks,
+  `em_call_finish` after the argument list, `ret_width`
+- `tests/compiler/args.n8`, `tests/compiler/args-test.s`
+- `scripts/build-args-test.ps1`, `scripts/verify-args-test.ps1`
+- `tests/regression/suites.ps1`, `docs/decisions.md` (D156),
+  `docs/architecture.md` (7.13), `planning/epics-and-stories.md`
+- `planning/claims/E04-S08.md`
+
+## Acceptance evidence
+
+- [x] Arguments are staged while parsed and pushed in reverse: `parse_args`
+  parses each argument with `em_ctx` on and calls `em_arg_push`, which emits
+  `PHA` for a byte and `PHA`/`TXA`/`PHA` for a word; at `)` `em_args_done`
+  pops the stages in reverse and pushes each through `frame_push_byte` or
+  `frame_push_word`. In the fixture `Calc.scale(n as byte)` reads `n` at
+  `FP+2` and returns 6 for `self.scale(3)`, so the value arrives at the
+  declared slot.
+- [x] The prologue is patched after the body is parsed: `em_prologue_start`
+  reserves `LDA #locals-lo`, `LDX #locals-hi`, `JSR frame_enter`, `BCC +1`,
+  `RTS` (10 bytes); `em_end_func` patches the local count from `em_locals`.
+  A refused frame executes the `RTS` and never runs `frame_leave` or the body.
+  The sizing pass counts the same 10 bytes because every byte goes through
+  `cg_put`.
+- [x] Returns leave the frame and preserve the result: `ret_width` maps the
+  current method's result type to 0/1/2 and `em_return` selects `fl_ret`,
+  `fl_ret_byte` or `fl_ret_word`; `fl_call`-style result saving keeps A (or
+  A/X) across the argument discard. The fixture's `twice` returns
+  `self.scale(3)` and the observed result is 6.
+- [x] Frames are opt-in: `parse_file_framed` sets `em_frame_want`, which
+  `em_reset` copies to `em_frame_on`; `parse_file`, `parse_file_base` and
+  `parse_file_sized` clear the request. A fixture that does not set it emits
+  no prologue.
+- [x] A test drives source with a parameter and a result in `x128`:
+  `scripts/verify-args-test.ps1` builds the compiler with `frames.s`, parses
+  `ARGS.N8`, runs `Calc.twice` with `JSR cg_code`, and requires result 6 with
+  no diagnostics, emitter, codegen or flow error. The suite reports PASS
+  through `tests/regression/run-regression.ps1 -Filter args` on a fresh build.
+
+## Coordination notes
+
+- Depends on E04-S07 (pass-3 addresses and routine sizes); both the count and
+  emit passes see the same prologue and argument bytes, so sizing stays
+  consistent.
+- E06-S07 remains blocked on interface arguments and four-byte returns; this
+  story covers byte and word parameters/results only.
+- `em_call` no longer emits on its own: it records the member row and result
+  width, and `em_call_finish` emits after the argument list. Parser call sites
+  were updated together with the emitter, so concrete calls still target the
+  same method labels.
+
+## Work log
+
+- Reused the existing `fl_call`/`fl_ret_*` encodings rather than adding a
+  second convention; only the push cells (`fl_push`, `fl_push_word`) are new.
+- The prologue initially emitted a manual `$20` opcode *and* `cg_jsr_abs`,
+  which writes its own `$20`; the resulting JSR target was garbage and the
+  fixture hung in the emitted code. Removing the manual opcode fixed it; the
+  timeout marker (`$C0` in the mailbox) is what identified the hang as
+  runtime rather than compile-time.
+- The first prologue used a five-byte flow branch to a failure label. It was
+  replaced with an inline `BCC +1`/`RTS`, which removes a label from the
+  routine and shrinks the prologue to 10 bytes.
+- `return` expressions and arguments never enabled `em_ctx`, so no value bytes
+  were emitted and the argument width came from stale state. The parser now
+  brackets both with `em_stmt_begin`/`em_stmt_end`.
+- The frame opt-in was first inferred from `fl_enter`, which lives in
+  uninitialized BSS and made a fixture without frames emit a prologue; it is
+  now an explicit `parse_file_framed` request.
+
+## Completion
+
+- Tests run: `tests/regression/run-regression.ps1 -Filter args` (fresh D81 and
+  fresh object files, run twice); `scripts/verify-args-test.ps1` standalone;
+  `git diff --check`; before the story, `-Filter driver` (1 passed).
+- Results: the args suite reports `args pass` with result 6 and no emitter,
+  codegen or flow error; emitted code is 0x78 bytes.
+- Files changed: `src/common/emit.inc`, `src/compiler/emit.s`,
+  `src/compiler/flow.s`, `src/compiler/parser.s`, `tests/compiler/args.n8`,
+  `tests/compiler/args-test.s`, `scripts/build-args-test.ps1`,
+  `scripts/verify-args-test.ps1`, `tests/regression/suites.ps1`,
+  `docs/decisions.md`, `docs/architecture.md`,
+  `planning/epics-and-stories.md`, `planning/claims/E04-S08.md`.
+- Remaining risks: interface arguments and four-byte interface returns are not
+  emitted (E06-S07); arguments are limited to `EM_ARG_MAX` (8) staged values;
+  the prologue's failure path returns without a diagnostic, so a refused frame
+  is currently silent; and the pre-existing `calls` suite failure described
+  below is unrelated but still red.
+
+## Known unrelated failure
+
+`calls` fails on a **fresh** build with `EM_ERR_LABEL` (7) and
+`FL_ERR_STATE` (3), `cg_len = 0x31`, while the same test passes when the build
+directory's older object files are reused. The identical signature reproduces
+on pristine `8a2d8ce` (verified by `git stash` + fresh `build/calls-test`), so
+it predates this story and points at emitter label state that is not
+deterministically initialized. It is not caused by E04-S08 and needs its own
+fix (initialize or validate the per-row method-label table before pass 2).
diff --git a/newLang/planning/epics-and-stories.md b/newLang/planning/epics-and-stories.md
index e0aaae5..9c0167b 100644
--- a/newLang/planning/epics-and-stories.md
+++ b/newLang/planning/epics-and-stories.md
@@ -412,6 +412,27 @@ Primary files: `sizing.s`, `codegen.s`, `output.s`, `src/runtime/startup.s`,
 
 ---
 
+### E04-S08 — Calling-convention emission (P0)
+
+**Depends on:** E04-S07
+
+**Acceptance criteria:**
+
+- A call's byte and word arguments are staged while the argument list is
+  parsed and pushed in reverse, so arg0 lands at `FP+2` in the callee and the
+  declared parameter slots read the passed values.
+- Every body gets a frame prologue whose local byte count is patched after the
+  body is parsed; a refused frame returns without running `frame_leave`, and
+  the sizing pass measures the same prologue bytes.
+- `return` leaves the frame and preserves a 0-, 1- or 2-byte result; the caller
+  keeps that result across the argument discard.
+- Frames are opt-in, so a fixture that does not request them emits no prologue
+  and keeps its previous code shape.
+- A test drives source with a parameter and a result in `x128` and checks the
+  returned value.
+
+---
+
 ## E05 — Classes, Heap, and Concrete Methods
 
 Purpose: implement object identity, encapsulated state, construction, and
diff --git a/newLang/scripts/build-args-test.ps1 b/newLang/scripts/build-args-test.ps1
new file mode 100644
index 0000000..ff07aa8
--- /dev/null
+++ b/newLang/scripts/build-args-test.ps1
@@ -0,0 +1,37 @@
+# E04-S08: build the argument/prologue/return fixture and a D81 with ARGS.N8.
+[CmdletBinding()]
+param()
+
+$ErrorActionPreference = "Stop"
+$project = Split-Path $PSScriptRoot -Parent
+Import-Module (Join-Path $PSScriptRoot "lib\Nova128Test.psm1") -Force
+$out = Join-Path $project "build\args-test"
+New-Item -ItemType Directory -Force -Path $out | Out-Null
+
+$link = Invoke-Nova128Link -Name "args-test" -BuildDir $out `
+    -Config (Join-Path $project "cfg\compiler.cfg") -Sources @(
+        (Join-Path $project "src\compiler\mmu.s"),
+        (Join-Path $project "src\compiler\diskio.s"),
+        (Join-Path $project "src\compiler\source.s"),
+        (Join-Path $project "src\compiler\lexer.s"),
+        (Join-Path $project "src\compiler\names.s"),
+        (Join-Path $project "src\compiler\diagnostics.s"),
+        (Join-Path $project "src\compiler\parser.s"),
+        (Join-Path $project "src\compiler\interfaces.s"),
+        (Join-Path $project "src\compiler\types.s"),
+        (Join-Path $project "src\compiler\codegen.s"),
+        (Join-Path $project "src\compiler\flow.s"),
+        (Join-Path $project "src\compiler\members.s"),
+        (Join-Path $project "src\compiler\sizing.s"),
+        (Join-Path $project "src\compiler\emit.s"),
+        (Join-Path $project "src\runtime\startup.s"),
+        (Join-Path $project "src\runtime\frames.s"),
+        (Join-Path $project "tests\compiler\args-test.s"))
+if ((Get-Nova128Label $link.Labels "test_entry") -ne 0x1C0D) { throw "test_entry is not at `$1C0D" }
+
+$pet = Join-Path $out "args.pet"
+[IO.File]::WriteAllBytes($pet, (ConvertTo-Nova128Petscii (Get-Content -Raw -LiteralPath (Join-Path $project "tests\compiler\args.n8"))))
+New-Nova128Disk -Path (Join-Path $out "args-test.d81") -Files @(
+    @{ Name = "args-test"; Path = $link.Program; Type = "prg" },
+    @{ Name = "args.n8"; Path = $pet; Type = "seq" })
+Write-Host ("built {0}" -f $link.Program)
diff --git a/newLang/scripts/verify-args-test.ps1 b/newLang/scripts/verify-args-test.ps1
new file mode 100644
index 0000000..72193e0
--- /dev/null
+++ b/newLang/scripts/verify-args-test.ps1
@@ -0,0 +1,37 @@
+# E04-S08: parse ARGS.N8, run Calc.twice, and check the byte result is 6.
+[CmdletBinding()]
+param([int]$Port = 6563, [int]$TimeoutSeconds = 180)
+
+$ErrorActionPreference = "Stop"
+$project = Split-Path $PSScriptRoot -Parent
+Import-Module (Join-Path $PSScriptRoot "lib\Nova128Test.psm1") -Force
+& (Join-Path $PSScriptRoot "build-args-test.ps1")
+
+$out = Join-Path $project "build\args-test"
+$disk = Join-Path $out "args-test.d81"
+$failures = @()
+
+$session = Start-Nova128X128 -Autostart $disk -Drive8Image $disk -Port $Port `
+    -Screenshot (Join-Path $out "args-vicii.png")
+try {
+    $mb = Wait-Nova128Mailbox $session -TimeoutSeconds $TimeoutSeconds
+} catch {
+    $dump = Read-Nova128Memory $session 0x1BF0 16
+    Write-Host ("timeout mailbox: " + (($dump | ForEach-Object { $_.ToString('X2') }) -join ' '))
+    throw
+} finally {
+    Stop-Nova128X128 $session
+}
+Write-Host ("mailbox: " + (($mb | ForEach-Object { $_.ToString('X2') }) -join ' '))
+
+if ($mb[2] -ne 0x02) { $failures += "producer $($mb[2]) expected 2 (compiler)" }
+if ($mb[3] -ne 0xA5) { $failures += ("fixture status {0:X2} failed at stage {1}" -f $mb[3], $mb[4]) }
+if ($mb[8] -ne 0) { $failures += "diagnostics present: $($mb[8])" }
+if ($mb[9] -ne 0) { $failures += "emitter error $($mb[9])" }
+if ($mb[10] -ne 6) { $failures += "result = $($mb[10]), expected 6" }
+if ($mb[12] -ne 0) { $failures += "codegen error $($mb[12])" }
+if ($mb[13] -ne 0) { $failures += "flow error $($mb[13])" }
+
+if ($failures) { throw ("args verification failed:`n  " + ($failures -join "`n  ")) }
+Write-Host ("emitted {0} bytes; Calc.twice returned {1}" -f $mb[11], $mb[10])
+Write-Host "verified byte argument push, sized prologue, and byte return in x128"
diff --git a/newLang/src/common/emit.inc b/newLang/src/common/emit.inc
index eb736f0..a1c5bef 100644
--- a/newLang/src/common/emit.inc
+++ b/newLang/src/common/emit.inc
@@ -19,10 +19,13 @@ EM_ERR_STATE    = 4             ; missing runtime entry or bad emitter state
 EM_ERR_FLOW     = 5             ; flow.s rejected a label or fixup
 EM_ERR_CALL     = 6             ; call shape not emitted yet
 EM_ERR_LABEL    = 7             ; no control-flow label available
+EM_ERR_ARG      = 8             ; argument width or staging limit
 
 EM_KIND_LOCAL   = 0
 EM_KIND_PARAM   = 1
 EM_MAX_ROUTINES = 24
+EM_ARG_MAX      = 8             ; arguments staged per call
+EM_PRO_SIZE     = 10            ; patched frame prologue bytes
 
 EM_LAST_LOCAL   = 1
 EM_LAST_SELF    = 2
diff --git a/newLang/src/compiler/emit.s b/newLang/src/compiler/emit.s
index 41734f5..171d615 100644
--- a/newLang/src/compiler/emit.s
+++ b/newLang/src/compiler/emit.s
@@ -31,8 +31,10 @@
 .import cg_init, cg_put, cg_lda_imm, cg_ldx_imm, cg_ldy_imm, cg_lda_zp, cg_lda_zpy, cg_sta_zpy
 .import cg_sta_zp, cg_adc_zp, cg_sbc_zp, cg_cmp_zp, cg_adc_imm, cg_clc, cg_sec, cg_jsr_abs, cg_rts
 .import gen_text, gen_text_len
-.import cg_len
+.import cg_len, cg_code
 .import fl_init, fl_set_base, fl_alloc, fl_label, fl_jmp, fl_jsr, fl_branch, fl_release, fl_finish, fl_err
+.import fl_enter, fl_leave, fl_discard, fl_push, fl_push_word
+.import fl_ret, fl_ret_byte, fl_ret_word
 .import mb_init, mb_self_store, mb_self_load, mb_ld_field, mb_st_field
 .import tb_get
 .import sc_base, sc_val, sc_len
@@ -46,7 +48,8 @@
 .export em_finish, em_self, em_member_receiver, em_field_ref, em_call
 .export em_open_block, em_cond_false, em_else, em_close_block
 .export em_for_var, em_for_save_end, em_for_save_step, em_for_default_step, em_for_setup
-.export em_return, em_method_begin, em_program_begin, em_entry
+.export em_return, em_method_begin, em_program_begin
+.export em_prologue_start, em_arg_push, em_args_done, em_call_finish, em_frame_want, em_entry
 .export em_pr_text: abs, em_pr_byte: abs, em_pr_cr: abs
 
 .assert EM_TMP + 1 < MB_SELF, error, "EM_TMP overlaps MB_SELF"
@@ -70,6 +73,8 @@ em_reset:
         lda #$00
 @init_flow:
         jsr fl_init
+        lda em_frame_want
+        sta em_frame_on
         lda em_base_set
         beq @base_done
         lda em_code_base
@@ -125,6 +130,9 @@ em_begin_func:
         sta em_last_kind
         sta em_tgt
         sta em_tgt_kind
+        sta em_pro_on
+        sta em_arg_n
+        sta em_call_pending
         rts
 
 ; The program body starts here. Its offset lets a fixture run the program
@@ -132,6 +140,42 @@ em_begin_func:
 em_program_begin:
         lda cg_len
         sta em_entry
+        jmp em_prologue_start
+
+; Reserve the patched frame prologue when the caller supplied frame_enter.
+; Layout: LDA #locals-lo, LDX #locals-hi, JSR fl_enter, BCC +1 (skip RTS).
+; A refused frame runs the RTS and returns without frame_leave; the body
+; never starts. em_end_func only patches the local count (EM_PRO_SIZE bytes).
+em_prologue_start:
+        lda em_pro_on
+        bne @done
+        lda em_frame_on
+        beq @done
+        lda fl_enter
+        ora fl_enter+1
+        beq @done
+        lda #$01
+        sta em_pro_on
+        lda cg_len
+        sta em_pro_at
+        lda #$A9                       ; LDA #locals-lo
+        jsr cg_put
+        lda #$00
+        jsr cg_put
+        lda #$A2                       ; LDX #locals-hi
+        jsr cg_put
+        lda #$00
+        jsr cg_put
+        lda fl_enter
+        ldx fl_enter+1
+        jsr cg_jsr_abs
+        lda #$90                       ; BCC +1
+        jsr cg_put
+        lda #$01
+        jsr cg_put
+        lda #$60                       ; RTS (refused frame)
+        jmp cg_put
+@done:  clc
         rts
 
 ; A signature starts: parameters begin at FP+2.
@@ -563,12 +607,21 @@ em_tramp_text:  jmp (em_pr_text)
 em_tramp_byte:  jmp (em_pr_byte)
 em_tramp_cr:    jmp (em_pr_cr)
 
-; End of a body: return to the caller.
+; End of a body: leave the frame and return to the caller.
 em_end_func:
         lda em_on
         beq @done
+        lda em_pro_on
+        beq @plain
+        jsr fl_ret
+        bcs @fail
+        jsr em_patch_pro
         jsr cg_rts
         bcs @fail
+        jmp @record
+@plain: jsr cg_rts
+        bcs @fail
+@record:
         ldx em_n_routines
         cpx #EM_MAX_ROUTINES
         bcs @done
@@ -581,11 +634,35 @@ em_end_func:
         rts
 @fail:  jmp em_fail
 
-; Explicit `return`: no frame prologue is emitted yet, so this is the exit.
+; Patch the prologue's local count. The failure branch is a flow fixup, so
+; fl_finish handles short and long distances; counting mode writes nothing.
+em_patch_pro:
+        lda em_count
+        bne @done
+        ldy em_pro_at
+        lda em_locals
+        sta cg_code+1,y
+        lda em_locals+1
+        sta cg_code+3,y
+@done:  clc
+        rts
+
+; A = result width (0, 1 or 2). Emit the matching exit.
 em_return:
+        sta em_ret_w
         lda em_on
         beq @done
-        jmp cg_rts
+        lda em_pro_on
+        beq @plain
+        lda em_ret_w
+        cmp #$01
+        beq @byte
+        cmp #$02
+        beq @word
+        jmp fl_ret
+@byte:  jmp fl_ret_byte
+@word:  jmp fl_ret_word
+@plain: jmp cg_rts
 @done:  clc
         rts
 
@@ -596,7 +673,8 @@ em_method_begin:
         txa
         jsr fl_label
         bcs @bad
-        jmp mb_self_store
+        jsr mb_self_store
+        jmp em_prologue_start
 @bad:   lda #EM_ERR_LABEL
         sta em_bad
         clc
@@ -625,24 +703,15 @@ em_method_label:
         sec
         rts
 
-; A = member row. Emit a zero-argument call through self.
+; A = member row. Record the call; the arguments are parsed next and
+; em_call_finish emits the self load, JSR and discard once they are pushed.
 em_call:
-        sta em_nid
-        jsr em_method_label
-        bcs @bad
-        stx em_lab
-        ldy em_nid
-        lda #C_MEM_PCOUNT
-        jsr tb_get
-        bne @call_bad
-        lda em_recv_self
-        beq @call_bad
-        jsr mb_self_load
-        bcs @fail
-        lda em_lab
-        jsr fl_jsr
-        bcs @fail
-        ldy em_nid
+        sta em_call_row
+        lda #$00
+        sta em_call_res
+        lda #$01
+        sta em_call_pending
+        ldy em_call_row
         lda #C_MEM_TBASE
         jsr tb_get
         cmp #TY_NONE
@@ -655,20 +724,174 @@ em_call:
         beq @byte
         cmp #TY_CHAR
         beq @byte
+        cmp #TY_INTEGER
+        beq @word
+        cmp #TY_WORD
+        beq @word
         lda #EM_ERR_CALL
+        sta em_bad
+        rts
+@void:  rts
+@byte:  lda #$01
+        sta em_call_res
+        rts
+@word:  lda #$02
+        sta em_call_res
+        rts
+
+; The just-parsed argument is in A (one byte) or A/X (two). Stage it on the
+; hardware stack; em_args_done pops the stages in reverse so arg0 lands at
+; FP+2 in the callee frame.
+em_arg_push:
+        lda em_on
+        beq @done
+        lda em_size
+        beq @bad
+        cmp #$02
+        bcs @two
+        ldy em_arg_n
+        cpy #EM_ARG_MAX
+        bcs @full
+        sta em_arg_w,y
+        inc em_arg_n
+        lda #$48                       ; PHA
+        jmp cg_put
+@two:   cmp #$02
+        bne @bad
+        ldy em_arg_n
+        cpy #EM_ARG_MAX
+        bcs @full
+        sta em_arg_w,y
+        inc em_arg_n
+        lda #$48                       ; PHA low
+        jsr cg_put
+        lda #$8A                       ; TXA
+        jsr cg_put
+        lda #$48                       ; PHA high
+        jmp cg_put
+@bad:   lda #EM_ERR_ARG
         sta em_bad
         clc
         rts
-@void:  lda #$00
-        sta em_size
+@full:  lda #EM_ERR_CALL
+        sta em_bad
+@done:  clc
+        rts
+
+; Pop the staged arguments in reverse and push each through frame_push_byte
+; or frame_push_word, so the first argument ends at FP+2.
+em_args_done:
+        lda em_on
+        beq @done
+        lda em_arg_n
+        beq @done
+        lda fl_push
+        ora fl_push+1
+        beq @state
+        lda em_arg_n
+        sta em_arg_i
+@loop:  dec em_arg_i
+        ldx em_arg_i
+        lda em_arg_w,x
+        cmp #$02
+        beq @word
+        lda #$68                       ; PLA
+        jsr cg_put
+        lda fl_push
+        ldx fl_push+1
+        jsr cg_jsr_abs
+        jmp @next
+@word:  lda #$68                       ; PLA -> high
+        jsr cg_put
+        lda #$AA                       ; TAX
+        jsr cg_put
+        lda #$68                       ; PLA -> low
+        jsr cg_put
+        lda fl_push_word
+        ldx fl_push_word+1
+        jsr cg_jsr_abs
+@next:  lda em_arg_i
+        bne @loop
+        lda #$00
+        sta em_arg_n
+@done:  clc
+        rts
+@state: lda #EM_ERR_STATE
+        sta em_bad
         clc
         rts
-@byte:  lda #$01
+
+; Emit the recorded call: self load, JSR, argument discard, result preserved
+; across the discard. Sets em_size to the result width.
+em_call_finish:
+        lda em_on
+        bne :+
+        jmp @done
+:       lda em_call_pending
+        bne :+
+        jmp @done
+:       lda #$00
+        sta em_call_pending
+        jsr em_arg_total
+        sta em_call_bytes
+        jsr mb_self_load
+        bcc :+
+        jmp @fail
+:
+        lda em_call_row
+        jsr em_method_label
+        bcs @bad
+        stx em_lab
+        lda em_lab
+        jsr fl_jsr
+        bcs @fail
+        lda em_call_res
+        beq @discard
+        cmp #$01
+        beq @save_byte
+        lda #$48                       ; PHA low
+        jsr cg_put
+        lda #$8A                       ; TXA
+        jsr cg_put
+        lda #$48                       ; PHA high
+        jsr cg_put
+        jmp @discard
+@save_byte:
+        lda #$48                       ; PHA
+        jsr cg_put
+@discard:
+        lda em_call_bytes
+        beq @restore
+        lda fl_discard
+        ora fl_discard+1
+        beq @state
+        lda em_call_bytes
+        jsr cg_lda_imm
+        lda #$00
+        jsr cg_ldx_imm
+        lda fl_discard
+        ldx fl_discard+1
+        jsr cg_jsr_abs
+@restore:
+        lda em_call_res
+        beq @ok
+        cmp #$01
+        beq @rest_byte
+        lda #$68                       ; PLA -> high
+        jsr cg_put
+        lda #$AA                       ; TAX
+        jsr cg_put
+        lda #$68                       ; PLA -> low
+        jsr cg_put
+        jmp @ok
+@rest_byte:
+        lda #$68                       ; PLA
+        jsr cg_put
+@ok:    lda em_call_res
         sta em_size
         clc
         rts
-@call_bad:
-        lda #EM_ERR_CALL
+@state: lda #EM_ERR_STATE
         sta em_bad
         clc
         rts
@@ -676,8 +899,52 @@ em_call:
         sta em_bad
         clc
         rts
+@done:  clc
+        rts
 @fail:  jmp em_fail
 
+; Total bytes of the recorded call's declared parameters. A width the ABI
+; does not know sets EM_ERR_CALL and leaves the total as it stood.
+em_arg_total:
+        lda #$00
+        sta em_arg_bytes
+        ldy em_call_row
+        lda #C_MEM_PCOUNT
+        jsr tb_get
+        beq @done
+        sta em_arg_i
+        ldy em_call_row
+        lda #C_MEM_PFIRST
+        jsr tb_get
+        sta em_nid
+@loop:  ldy em_nid
+        lda #C_PAR_TBASE
+        jsr tb_get
+        cmp #TY_BOOLEAN
+        beq @one
+        cmp #TY_BYTE
+        beq @one
+        cmp #TY_CHAR
+        beq @one
+        cmp #TY_INTEGER
+        beq @two
+        cmp #TY_WORD
+        beq @two
+        lda #EM_ERR_CALL
+        sta em_bad
+        jmp @done
+@one:   lda #$01
+        jmp @add
+@two:   lda #$02
+@add:   clc
+        adc em_arg_bytes
+        sta em_arg_bytes
+        inc em_nid
+        dec em_arg_i
+        bne @loop
+@done:  lda em_arg_bytes
+        rts
+
 ; The next member is applied to self (a field or a call).
 em_member_receiver:
         lda #$00
@@ -1165,3 +1432,17 @@ em_count:       .res 1
 em_n_routines:  .res 1
 em_routine_start: .res 1
 em_routine_sizes: .res EM_MAX_ROUTINES
+em_pro_on:      .res 1
+em_pro_at:      .res 1
+em_ret_w:       .res 1
+em_arg_n:       .res 1
+em_arg_i:       .res 1
+em_arg_b:       .res 1
+em_arg_bytes:   .res 1
+em_arg_w:       .res EM_ARG_MAX
+em_call_row:    .res 1
+em_call_res:    .res 1
+em_call_bytes:  .res 1
+em_call_pending: .res 1
+em_frame_on:    .res 1
+em_frame_want:  .res 1
diff --git a/newLang/src/compiler/flow.s b/newLang/src/compiler/flow.s
index e548c7b..ddd8fe8 100644
--- a/newLang/src/compiler/flow.s
+++ b/newLang/src/compiler/flow.s
@@ -68,7 +68,7 @@
 .export fl_branch_true, fl_branch_false
 .export fl_call, fl_call_byte, fl_call_word
 .export fl_prologue, fl_ret, fl_ret_byte, fl_ret_word
-.export fl_err, fl_enter, fl_leave, fl_discard
+.export fl_err, fl_enter, fl_leave, fl_discard, fl_push, fl_push_word
 
 .segment "CODE"
 
@@ -762,4 +762,7 @@ fl_fix:         .res FL_FIX_MAX * FL_FIX_SIZE
 fl_enter:       .res 2
 fl_leave:       .res 2
 fl_discard:     .res 2
+fl_push:        .res 2
+fl_push_word:   .res 2
+
 fl_code_base:   .res 2
diff --git a/newLang/src/compiler/parser.s b/newLang/src/compiler/parser.s
index eb06456..ec6f52a 100644
--- a/newLang/src/compiler/parser.s
+++ b/newLang/src/compiler/parser.s
@@ -65,9 +65,11 @@
 .import em_open_block, em_cond_false, em_else, em_close_block
 .import em_for_var, em_for_save_end, em_for_save_step, em_for_default_step, em_for_setup
 .import em_return, em_method_begin, em_program_begin
+.import em_arg_push, em_args_done, em_call_finish, em_frame_want
 .import sz_capture, sz_verify
 
-.export parse_file, parse_file_base, parse_file_sized, tb_get, tb_set, mem_count, par_count
+.export parse_file, parse_file_base, parse_file_sized, parse_file_framed
+.export tb_get, tb_set, mem_count, par_count
 .export gen_text, gen_text_len, gen_print_count, gen_bad
 .export ex_trace_on, ex_min_sp
 .exportzp ex_tp
@@ -96,6 +98,15 @@ ex_tp:          .res 2             ; test trace pointer (ex_trace_on)
 .segment "CODE"
 
 parse_file:
+        lda #$00
+        sta em_frame_want
+        jmp parse_plain
+
+; Same as parse_file, but generated bodies get a frame_enter prologue.
+parse_file_framed:
+        lda #$01
+        sta em_frame_want
+parse_plain:
         lda #$00
         sta em_base_set
         sta em_count
@@ -107,6 +118,7 @@ parse_file:
 parse_file_base:
         jsr em_set_base
         lda #$00
+        sta em_frame_want
         sta em_count
         sta diag_suppress
         jmp parse_start
@@ -114,6 +126,8 @@ parse_file_base:
 ; Count pass, snapshot, then an emitting pass. The caller installs the final
 ; code base first. Diagnostics are reported only by the count pass.
 parse_file_sized:
+        lda #$00
+        sta em_frame_want
         lda #$01
         sta em_count
         lda #$00
@@ -1325,8 +1339,23 @@ parse_statement:
         beq @bare
         cmp #TOK_EOF
         beq @bare
+        lda pass
+        cmp #$02
+        bne @ret_parse
+        lda em_on
+        beq @ret_parse
+        jsr em_stmt_begin
+@ret_parse:
         jsr parse_full
+        lda pass
+        cmp #$02
+        bne @ret_typed
+        lda em_on
+        beq @ret_typed
+        jsr em_stmt_end
+@ret_typed:
         jsr check_return
+        jsr ret_width
         jsr em_return
         jmp end_stmt
 @bare:  lda ret_kind
@@ -1335,7 +1364,8 @@ parse_statement:
         lda #E_SEM_MISSING_RETURN
         ldx stmt_col
         jsr err2_col
-:       jsr em_return
+:       jsr ret_width
+        jsr em_return
         jmp end_stmt
 @else:  ldx depth                  ; only once, directly inside an if
         lda block_stack-1,x
@@ -1623,6 +1653,32 @@ check_return:
         jmp err2_col
 @ok:    rts
 
+; A = the current method's result width (0, 1 or 2) for em_return. A program
+; and a void method report 0, so their return leaves the frame without a value.
+ret_width:
+        lda cur_mem
+        cmp #$FF
+        beq @zero
+        ldy cur_mem
+        lda #C_MEM_TBASE
+        jsr tb_get
+        cmp #TY_BOOLEAN
+        beq @one
+        cmp #TY_BYTE
+        beq @one
+        cmp #TY_CHAR
+        beq @one
+        cmp #TY_INTEGER
+        beq @two
+        cmp #TY_WORD
+        beq @two
+@zero:  lda #$00
+        rts
+@one:   lda #$01
+        rts
+@two:   lda #$02
+        rts
+
 ; The for variable must be a byte, word, or integer. A bad one is reported
 ; as "FOR NOT ALLOWED ON <type>" and its type becomes unknown.
 check_for_var:
@@ -1985,6 +2041,13 @@ parse_postfix:
         jsr em_call
 :
         jsr parse_args
+        lda pass
+        cmp #$02
+        bne :+
+        lda em_on
+        beq :+
+        jsr em_call_finish
+:
         ldx ex_depth
         jsr load_sc
         lda #EX_CALL
@@ -2132,8 +2195,23 @@ parse_args:
 @arg:   lda tok_col
         ldx ex_depth
         sta ex_acol,x
+        lda pass
+        cmp #$02
+        bne @arg_parse
+        lda em_on
+        beq @arg_parse
+        jsr em_stmt_begin
+@arg_parse:
         jsr parse_full
         jsr check_arg
+        lda pass
+        cmp #$02
+        bne :+
+        lda em_on
+        beq :+
+        jsr em_stmt_end
+        jsr em_arg_push
+:
         lda tok_kind
         cmp #TOK_COMMA
         bne :+
@@ -2145,12 +2223,19 @@ parse_args:
         jmp syn_error
 @close: ldx ex_depth               ; too few arguments: reported at `)`
         lda ex_cchk,x
-        beq @done
+        beq @args_done
         lda ex_carg,x
         cmp ex_cpc,x
-        bcs @done
+        bcs @args_done
         lda tok_col
         jsr count_error
+@args_done:
+        lda pass
+        cmp #$02
+        bne @done
+        lda em_on
+        beq @done
+        jsr em_args_done
 @done:  jmp next
 
 ; Argument ex_carg (sc) against its parameter. The first argument beyond
diff --git a/newLang/tests/compiler/args-test.s b/newLang/tests/compiler/args-test.s
new file mode 100644
index 0000000..66b4db5
--- /dev/null
+++ b/newLang/tests/compiler/args-test.s
@@ -0,0 +1,179 @@
+; E04-S08: parse args.n8 and run Calc.twice, which pushes one byte argument,
+; enters a sized frame in Calc.scale, and returns a byte.
+;
+; The first emitted body is Calc.twice at cg_code+0. The fixture supplies the
+; frame runtime entry points, so the prologue emitted at the start of twice
+; calls frame_enter(0), and scale reads its parameter at FP+2. twice returns
+; self.scale(3); scale returns n + 3, so the expected result is 6.
+;
+; Mailbox data: +0 diagnostics, +1 em_bad, +2 result, +3 cg_len, +4 cg_err,
+; +5 fl_err. FAIL stage: 1 load, 2 diagnostics, 3 emitter, 4 result.
+
+.setcpu "6502"
+.include "../../src/common/c128.inc"
+.include "../../src/common/mailbox.inc"
+.include "../../src/common/formats.inc"
+
+.import mmu_init, mmu_bank0, mmu_shutdown
+.import src_load, src_dos_code, src_kernal_err
+.import names_init
+.import diag_init, diag_finish, diag_screen
+.import parse_file_framed
+.import em_on, em_bad
+.import cg_code, cg_len, cg_err
+.import fl_err, fl_enter, fl_leave, fl_discard, fl_push, fl_push_word
+.import frame_init, frame_enter, frame_leave, frame_discard
+.import frame_push_byte, frame_push_word
+
+.macro ON_ZERO target
+        beq :+
+        jmp target
+:
+.endmacro
+
+.segment "LOADADDR"
+        .word $1C01
+
+.segment "EXEHDR"
+        .word basic_end
+        .word 10
+        .byte $9E
+        .byte "7181"
+        .byte $00
+basic_end:
+        .word $0000
+
+.segment "MAILBOX"
+        .res MB_SIZE
+
+.segment "STARTUP"
+.export test_entry
+test_entry:
+        php
+        sei
+        ldx #MB_SIZE - 1
+        lda #$00
+@clr:   sta MB_BASE,x
+        dex
+        bpl @clr
+        lda #MB_MAGIC0_VALUE
+        sta MB_MAGIC0
+        lda #MB_MAGIC1_VALUE
+        sta MB_MAGIC1
+        lda #MB_PROD_COMPILER
+        sta MB_PRODUCER
+        lda #MB_ST_RUNNING
+        sta MB_STATUS
+
+        jsr mmu_init
+        jsr mmu_bank0
+
+        lda #<frame_enter
+        sta fl_enter
+        lda #>frame_enter
+        sta fl_enter+1
+        lda #<frame_leave
+        sta fl_leave
+        lda #>frame_leave
+        sta fl_leave+1
+        lda #<frame_discard
+        sta fl_discard
+        lda #>frame_discard
+        sta fl_discard+1
+        lda #<frame_push_byte
+        sta fl_push
+        lda #>frame_push_byte
+        sta fl_push+1
+        lda #<frame_push_word
+        sta fl_push_word
+        lda #>frame_push_word
+        sta fl_push_word+1
+
+        lda #<name_src
+        ldx #>name_src
+        ldy #name_src_len
+        jsr src_load
+        cmp #$00
+        ON_ZERO @fail_load
+
+        jsr names_init
+        lda #<name_src
+        ldx #>name_src
+        ldy #name_src_len
+        jsr diag_init
+        lda #$00
+        sta diag_screen
+        lda #$01
+        sta em_on
+
+        jsr parse_file_framed
+        jsr diag_finish
+        sta MB_DATA+0
+        lda em_bad
+        sta MB_DATA+1
+        lda cg_len
+        sta MB_DATA+3
+        lda cg_err
+        sta MB_DATA+4
+        lda fl_err
+        sta MB_DATA+5
+
+        lda MB_DATA+0
+        ON_ZERO @fail_diag
+        lda em_bad
+        ON_ZERO @fail_emit
+        lda cg_err
+        ON_ZERO @fail_cg
+        lda fl_err
+        ON_ZERO @fail_cg
+        jmp @run
+@fail_cg:
+        lda #3
+        jmp @fail
+
+@run:   jsr frame_init
+        ldx #3                       ; descriptor + zeroed byte field
+        lda #$00
+@obj:   sta object,x
+        dex
+        bpl @obj
+        lda #<object
+        ldx #>object
+        jsr cg_code
+        sta MB_DATA+2
+        cmp #$06
+        beq @pass
+        lda #4
+        jmp @fail
+
+@fail_load:
+        sta MB_DATA+4
+        lda src_dos_code
+        sta MB_DATA+5
+        lda src_kernal_err
+        sta MB_DATA+6
+        lda #1
+        jmp @fail
+@fail_diag:
+        lda #2
+        jmp @fail
+@fail_emit:
+        lda #3
+@fail:  sta MB_CODE
+        lda #$00
+        sta MB_CODE+1
+        lda #MB_ST_FAIL
+        sta MB_STATUS
+        jmp @done
+@pass:  lda #MB_ST_PASS
+        sta MB_STATUS
+@done:  jsr mmu_shutdown
+        plp
+        rts
+
+.segment "RODATA"
+name_src:       .byte "ARGS.N8"
+name_src_len = * - name_src
+
+.segment "BSS"
+object:         .res 4
diff --git a/newLang/tests/compiler/args.n8 b/newLang/tests/compiler/args.n8
new file mode 100644
index 0000000..7eb99f2
--- /dev/null
+++ b/newLang/tests/compiler/args.n8
@@ -0,0 +1,13 @@
+// E04-S08 calling-convention source: one byte argument and a byte result.
+class Calc
+  public method twice() as byte
+    return self.scale(3)
+  end method
+
+  public method scale(n as byte) as byte
+    return n + 3
+  end method
+end class
+
+program args
+end program
diff --git a/newLang/tests/regression/suites.ps1 b/newLang/tests/regression/suites.ps1
index 15618e2..1ec5bc5 100644
--- a/newLang/tests/regression/suites.ps1
+++ b/newLang/tests/regression/suites.ps1
@@ -16,6 +16,7 @@
     @{ Name = "print"; Script = "scripts\verify-print-test.ps1"; BuildDir = "print-test"; Timeout = 300 }
     @{ Name = "statements"; Script = "scripts\verify-statements-test.ps1"; BuildDir = "statements-test"; Timeout = 300 }
     @{ Name = "calls"; Script = "scripts\verify-calls-test.ps1"; BuildDir = "calls-test"; Timeout = 300 }
+    @{ Name = "args"; Script = "scripts\verify-args-test.ps1"; BuildDir = "args-test"; Timeout = 300; Success = "verified byte argument push" }
     @{ Name = "class"; Script = "scripts\verify-class-test.ps1"; BuildDir = "class-test"; Timeout = 300 }
     @{ Name = "member"; Script = "scripts\verify-member-test.ps1"; BuildDir = "member-test"; Timeout = 300 }
     @{ Name = "heap"; Script = "scripts\verify-heap-test.ps1"; BuildDir = "heap-test"; Timeout = 300 }