diff --git a/CHANGELOG.md b/CHANGELOG.md
index 183a8e41..ff7daa2d 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -13,6 +13,7 @@
 - `--jobs auto` / `-j auto` caps parallel concurrency at the CPU core count (portable across Linux/macOS/BSD); the default stays unlimited (#766)
 
 ### Changed
+- Faster test runs: each test's subshell result payload is emitted with `printf` (a builtin) instead of a `cat <<EOF` heredoc, removing one `cat` fork per test. No behaviour change
 - Faster cold start: the run-output scratch directory and the shared temp directory are created with a single `mkdir -p` call instead of one fork each (2 -> 1 `mkdir` forks per cold start). No behaviour change
 - Faster test runs: the runner detects a failed source (syntax error / unexpected EOF) and discovery detects the `.bash` pattern variant with shell `case` matches instead of `grep` forks (2 -> 0 `grep` forks per sourced test file). No behaviour change
 - Faster cold start: the base64 `-w` capability is detected with a shell `case` on `base64 --help` instead of piping into a `grep` fork (1 -> 0 `grep` forks per cold start). Same detection, no behaviour change
diff --git a/src/state.sh b/src/state.sh
index 52f91255..99a2b25f 100644
--- a/src/state.sh
+++ b/src/state.sh
@@ -262,7 +262,11 @@ function bashunit::state::export_subshell_context() {
   bashunit::state::encode_field "$_BASHUNIT_TEST_HOOK_MESSAGE"
   encoded_test_hook_message=$_BASHUNIT_STATE_ENCODED_OUT
 
-  cat <<EOF
+  # Emit the encoded result payload with `printf` (a builtin) instead of a
+  # `cat <<EOF` heredoc: this runs once per test, so avoiding the fork removes
+  # one process per test. The `\`-continued string keeps the per-field layout
+  # and produces the exact same single line the heredoc did.
+  local payload="\
 ##ASSERTIONS_FAILED=$_BASHUNIT_ASSERTIONS_FAILED\
 ##ASSERTIONS_PASSED=$_BASHUNIT_ASSERTIONS_PASSED\
 ##ASSERTIONS_SKIPPED=$_BASHUNIT_ASSERTIONS_SKIPPED\
@@ -272,8 +276,8 @@ function bashunit::state::export_subshell_context() {
 ##TEST_HOOK_FAILURE=$_BASHUNIT_TEST_HOOK_FAILURE\
 ##TEST_HOOK_MESSAGE=$encoded_test_hook_message\
 ##TEST_TITLE=$encoded_test_title\
-##TEST_OUTPUT=$encoded_test_output##
-EOF
+##TEST_OUTPUT=$encoded_test_output##"
+  printf '%s\n' "$payload"
 }
 
 function bashunit::state::calculate_total_assertions() {
diff --git a/tests/acceptance/bashunit_run_forks_test.sh b/tests/acceptance/bashunit_run_forks_test.sh
index 3e1c89f3..0cf367ad 100644
--- a/tests/acceptance/bashunit_run_forks_test.sh
+++ b/tests/acceptance/bashunit_run_forks_test.sh
@@ -26,3 +26,31 @@ function test_running_a_test_file_does_not_fork_grep() {
 
   assert_equals 0 "$grep_forks"
 }
+
+# Regression guard: exporting a test's subshell result must not fork `cat`. It
+# used to emit the encoded result payload with a `cat <<EOF` heredoc — one fork
+# per test — which `printf` (a builtin) does without forking. Four independent
+# tests must therefore fork `cat` far fewer than four times.
+function test_running_tests_does_not_fork_cat_per_test() {
+  if bashunit::check_os::is_windows; then
+    bashunit::skip "process tracing is unreliable under Git Bash" && return
+  fi
+
+  local dir
+  dir="$(bashunit::temp_dir)"
+  local fixture="$dir/cat_forks_test.sh"
+  {
+    echo 'function test_a() { assert_true true; }'
+    echo 'function test_b() { assert_true true; }'
+    echo 'function test_c() { assert_true true; }'
+    echo 'function test_d() { assert_true true; }'
+  } >"$fixture"
+
+  local trace
+  trace="$(PS4='+ ' bash -x ./bashunit --no-parallel "$fixture" 2>&1 >/dev/null)"
+
+  local cat_forks
+  cat_forks="$(printf '%s\n' "$trace" | grep -cE '^\++ +/?[a-z/]*cat( |$)' || true)"
+
+  assert_less_or_equal_than 1 "$cat_forks"
+}