diff --git a/test/helpers/bounded-child-output.test.ts b/test/helpers/bounded-child-output.test.ts
index 32c0e15e21b0..2ec1b29656ce 100644
--- a/test/helpers/bounded-child-output.test.ts
+++ b/test/helpers/bounded-child-output.test.ts
@@ -22,4 +22,29 @@ describe("bounded child output", () => {
 
     expect(output.text()).toBe("x-recent");
   });
+
+  it("drops split UTF-8 prefixes after buffered output overflow", () => {
+    const output = createBoundedChildOutput(7);
+
+    output.append(Buffer.from("prefix 😀"));
+    output.append(Buffer.from("tail"));
+
+    expect(output.text()).toBe("tail");
+  });
+
+  it("drops split UTF-8 prefixes from single oversized chunks", () => {
+    const output = createBoundedChildOutput(7);
+
+    output.append(Buffer.from("prefix 😀tail"));
+
+    expect(output.text()).toBe("tail");
+  });
+
+  it("preserves replacement output for incomplete trailing bytes", () => {
+    const output = createBoundedChildOutput(7);
+
+    output.append(Buffer.from([0x61, 0xe2]));
+
+    expect(output.text()).toBe("a�");
+  });
 });
diff --git a/test/helpers/bounded-child-output.ts b/test/helpers/bounded-child-output.ts
index d4669e710c81..d2826f5078f2 100644
--- a/test/helpers/bounded-child-output.ts
+++ b/test/helpers/bounded-child-output.ts
@@ -1,5 +1,15 @@
+import { StringDecoder } from "node:string_decoder";
+
 export const DEFAULT_CHILD_OUTPUT_TAIL_BYTES = 128 * 1024;
 
+function decodeUtf8Tail(buffer: Buffer): string {
+  let start = 0;
+  while (start < buffer.length && (buffer[start]! & 0b1100_0000) === 0b1000_0000) {
+    start += 1;
+  }
+  return new StringDecoder("utf8").end(buffer.subarray(start));
+}
+
 export function createBoundedChildOutput(maxBytes = DEFAULT_CHILD_OUTPUT_TAIL_BYTES) {
   const limit =
     Number.isInteger(maxBytes) && maxBytes > 0 ? maxBytes : DEFAULT_CHILD_OUTPUT_TAIL_BYTES;
@@ -37,7 +47,7 @@ export function createBoundedChildOutput(maxBytes = DEFAULT_CHILD_OUTPUT_TAIL_BY
       trim();
     },
     text(): string {
-      return Buffer.concat(chunks, totalBytes).toString("utf8");
+      return decodeUtf8Tail(Buffer.concat(chunks, totalBytes));
     },
   };
 }