diff --git a/FileAPI/unicode.html b/FileAPI/unicode.html
new file mode 100644
index 0000000000..ce3e3579d7
--- /dev/null
+++ b/FileAPI/unicode.html
@@ -0,0 +1,46 @@
+<!DOCTYPE html>
+<meta charset="utf-8">
+<title>Blob/Unicode interaction: normalization and encoding</title>
+<script src="/resources/testharness.js"></script>
+<script src="/resources/testharnessreport.js"></script>
+<script>
+'use strict';
+
+const OMICRON_WITH_OXIA = '\u1F79'; // NFC normalized to U+3CC
+const CONTAINS_UNPAIRED_SURROGATES = 'abc\uDC00def\uD800ghi';
+const REPLACED = 'abc\uFFFDdef\uFFFDghi';
+
+function readBlobAsPromise(blob) {
+  return new Promise((resolve, reject) => {
+    const reader = new FileReader();
+    reader.readAsText(blob);
+    reader.onload = () => resolve(reader.result);
+    reader.onerror = () => reject(reader.error);
+  });
+}
+
+promise_test(async t => {
+  const blob = new Blob([OMICRON_WITH_OXIA]);
+  const result = await readBlobAsPromise(blob);
+  assert_equals(result, OMICRON_WITH_OXIA, 'String should not be normalized');
+}, 'Test that strings are not NFC normalized by Blob constructor');
+
+promise_test(async t => {
+  const file = new File([OMICRON_WITH_OXIA], 'name');
+  const result = await readBlobAsPromise(file);
+  assert_equals(result, OMICRON_WITH_OXIA, 'String should not be normalized');
+}, 'Test that strings are not NFC normalized by File constructor');
+
+promise_test(async t => {
+  const blob = new Blob([CONTAINS_UNPAIRED_SURROGATES]);
+  const result = await readBlobAsPromise(blob);
+  assert_equals(result, REPLACED, 'Unpaired surrogates should be replaced.');
+}, 'Test that unpaired surrogates are replaced by Blob constructor');
+
+promise_test(async t => {
+  const file = new File([CONTAINS_UNPAIRED_SURROGATES], 'name');
+  const result = await readBlobAsPromise(file);
+  assert_equals(result, REPLACED, 'Unpaired surrogates should be replaced.');
+}, 'Test that unpaired surrogates are replaced by File constructor');
+
+</script>
