diff --git a/README.md b/README.md
index 972f3c69..c5962007 100644
--- a/README.md
+++ b/README.md
@@ -154,13 +154,44 @@ import { utf16fromString, utf16toString } from '@exodus/bytes/utf16.js'
 import { utf16fromStringLoose, utf16toStringLoose } from '@exodus/bytes/utf16.js'
 ```
 
-_These methods by design encode/decode BOM (codepoint `U+FEFF` Byte Order Mark) as-is._
+_These methods by design encode/decode BOM (codepoint `U+FEFF` Byte Order Mark) as-is._\
 _If you need BOM handling or detection, use `@exodus/bytes/encoding.js`_
 
 #### `utf16fromString(string, format = 'uint16')`
+
+Encode a string to UTF-16 bytes (strict mode)
+
+Throws on invalid Unicode (unpaired surrogates)
+
 #### `utf16fromStringLoose(string, format = 'uint16')`
-#### `utf16toString(arr, 'uint16')`
-#### `utf16toStringLoose(arr, 'uint16')`
+
+Encode a string to UTF-16 bytes (loose mode)
+
+Replaces invalid Unicode (unpaired surrogates) with replacement codepoints `U+FFFD`
+per [WHATWG Encoding](https://encoding.spec.whatwg.org/) specification.
+
+_Such replacement is a non-injective function, is irreversible and causes collisions.\
+Prefer using strict throwing methods for cryptography applications._
+
+#### `utf16toString(arr, format = 'uint16')`
+
+Decode UTF-16 bytes to a string (strict mode)
+
+Throws on invalid UTF-16 byte sequences
+
+Throws on non-even byte length.
+
+#### `utf16toStringLoose(arr, format = 'uint16')`
+
+Decode UTF-16 bytes to a string (loose mode)
+
+Replaces invalid UTF-16 byte sequences with replacement codepoints `U+FFFD`
+per [WHATWG Encoding](https://encoding.spec.whatwg.org/) specification.
+
+_Such replacement is a non-injective function, is irreversible and causes collisions.\
+Prefer using strict throwing methods for cryptography applications._
+
+Throws on non-even byte length.
 
 ### `@exodus/bytes/single-byte.js`
 
diff --git a/array.d.ts b/array.d.ts
index 81dc817b..c299a76a 100644
--- a/array.d.ts
+++ b/array.d.ts
@@ -15,6 +15,7 @@
 // < TypeScript 5.7 doesn't support templates for Uint8Array.
 // So this type is defined as a workaround to evaluate to Uint8Array<ArrayBuffer> on all versions of TypeScript.
 export type Uint8ArrayBuffer = ReturnType<typeof Uint8Array.from>;
+export type Uint16ArrayBuffer = ReturnType<typeof Uint16Array.from>;
 
 /**
  * Output format for typed array conversions
diff --git a/package.json b/package.json
index bcb87dd2..854500d7 100644
--- a/package.json
+++ b/package.json
@@ -110,6 +110,7 @@
     "/single-byte.d.ts",
     "/single-byte.node.js",
     "/utf16.js",
+    "/utf16.d.ts",
     "/utf16.node.js",
     "/utf8.js",
     "/utf8.d.ts",
@@ -176,6 +177,7 @@
       "default": "./encoding-browser.js"
     },
     "./utf16.js": {
+      "types": "./utf16.d.ts",
       "node": "./utf16.node.js",
       "default": "./utf16.js"
     },
diff --git a/utf16.d.ts b/utf16.d.ts
new file mode 100644
index 00000000..62550f5f
--- /dev/null
+++ b/utf16.d.ts
@@ -0,0 +1,92 @@
+/**
+ * UTF-16 encoding/decoding
+ *
+ * ```js
+ * import { utf16fromString, utf16toString } from '@exodus/bytes/utf16.js'
+ *
+ * // loose
+ * import { utf16fromStringLoose, utf16toStringLoose } from '@exodus/bytes/utf16.js'
+ * ```
+ *
+ * _These methods by design encode/decode BOM (codepoint `U+FEFF` Byte Order Mark) as-is._\
+ * _If you need BOM handling or detection, use `@exodus/bytes/encoding.js`_
+ *
+ * @module @exodus/bytes/utf16.js
+ */
+
+/// <reference types="node" />
+
+import type { Uint8ArrayBuffer, Uint16ArrayBuffer } from './array.js';
+
+/**
+ * Output format for UTF-16 encoding
+ */
+export type Utf16Format = 'uint16' | 'uint8-le' | 'uint8-be';
+
+/**
+ * Encode a string to UTF-16 bytes (strict mode)
+ *
+ * Throws on invalid Unicode (unpaired surrogates)
+ *
+ * @param string - The string to encode
+ * @param format - Output format (default: 'uint16')
+ * @returns The encoded bytes
+ */
+export function utf16fromString(string: string, format?: 'uint16'): Uint16ArrayBuffer;
+export function utf16fromString(string: string, format: 'uint8-le'): Uint8ArrayBuffer;
+export function utf16fromString(string: string, format: 'uint8-be'): Uint8ArrayBuffer;
+export function utf16fromString(string: string, format?: Utf16Format): Uint16ArrayBuffer | Uint8ArrayBuffer;
+
+/**
+ * Encode a string to UTF-16 bytes (loose mode)
+ *
+ * Replaces invalid Unicode (unpaired surrogates) with replacement codepoints `U+FFFD`
+ * per [WHATWG Encoding](https://encoding.spec.whatwg.org/) specification.
+ *
+ * _Such replacement is a non-injective function, is irreversible and causes collisions.\
+ * Prefer using strict throwing methods for cryptography applications._
+ *
+ * @param string - The string to encode
+ * @param format - Output format (default: 'uint16')
+ * @returns The encoded bytes
+ */
+export function utf16fromStringLoose(string: string, format?: 'uint16'): Uint16ArrayBuffer;
+export function utf16fromStringLoose(string: string, format: 'uint8-le'): Uint8ArrayBuffer;
+export function utf16fromStringLoose(string: string, format: 'uint8-be'): Uint8ArrayBuffer;
+export function utf16fromStringLoose(string: string, format?: Utf16Format): Uint16ArrayBuffer | Uint8ArrayBuffer;
+
+/**
+ * Decode UTF-16 bytes to a string (strict mode)
+ *
+ * Throws on invalid UTF-16 byte sequences
+ *
+ * Throws on non-even byte length.
+ *
+ * @param arr - The bytes to decode
+ * @param format - Input format (default: 'uint16')
+ * @returns The decoded string
+ */
+export function utf16toString(arr: Uint16ArrayBuffer, format?: 'uint16'): string;
+export function utf16toString(arr: Uint8ArrayBuffer, format: 'uint8-le'): string;
+export function utf16toString(arr: Uint8ArrayBuffer, format: 'uint8-be'): string;
+export function utf16toString(arr: Uint16ArrayBuffer | Uint8ArrayBuffer, format?: Utf16Format): string;
+
+/**
+ * Decode UTF-16 bytes to a string (loose mode)
+ *
+ * Replaces invalid UTF-16 byte sequences with replacement codepoints `U+FFFD`
+ * per [WHATWG Encoding](https://encoding.spec.whatwg.org/) specification.
+ *
+ * _Such replacement is a non-injective function, is irreversible and causes collisions.\
+ * Prefer using strict throwing methods for cryptography applications._
+ *
+ * Throws on non-even byte length.
+ *
+ * @param arr - The bytes to decode
+ * @param format - Input format (default: 'uint16')
+ * @returns The decoded string
+ */
+export function utf16toStringLoose(arr: Uint16ArrayBuffer, format?: 'uint16'): string;
+export function utf16toStringLoose(arr: Uint8ArrayBuffer, format: 'uint8-le'): string;
+export function utf16toStringLoose(arr: Uint8ArrayBuffer, format: 'uint8-be'): string;
+export function utf16toStringLoose(arr: Uint16ArrayBuffer | Uint8ArrayBuffer, format?: Utf16Format): string;