Base64 Decode Online — Free, Instant, Browser-Only

What is Base64 Decoding?

Base64 encoding turns arbitrary binary data (bytes) into a string of 64 printable ASCII characters — A-Z, a-z, 0-9, +, and /. Decoding reverses that: given a Base64 string, you recover the original bytes. Those bytes might be UTF-8 text, JSON, a JWT header/payload, an image, a PDF, or any other binary blob.

Base64 is used everywhere: HTTP Basic Authentication headers, JWT tokens, data URLs in CSS/HTML, MIME email attachments, SSH keys, TLS certificates, and countless serialization protocols. Whenever you need to move binary data through a text-only channel, Base64 is the go-to encoding.

Why Use an Online Base64 Decoder Instead of the Command Line?

You could just run echo "SGVsbG8=" | base64 -d in your terminal — and that works fine for small ASCII strings. But there are real cases where a browser tool wins:

• Long strings — pasting a 2000-character JWT into a terminal is painful. In a browser, ctrl+V just works.
• UTF-8 gotchas — the CLI base64 command decodes to raw bytes and dumps them to stdout. Non-ASCII characters appear mangled. Our tool uses TextDecoder to render UTF-8 correctly.
• URL-safe variants — the CLI doesn't handle -/_ substitutions. You'd have to tr them back to +// first.
• Missing padding — JWT payloads have padding stripped. CLI base64 -d errors out; our tool auto-adds the padding.
• Privacy — same as CLI, nothing leaves your machine. Better than pasting a token into a random third-party server.

How Base64 Decoding Works Under the Hood

Base64 groups 3 bytes of input into 4 output characters. Each output character represents 6 bits (2⁶ = 64 possible values, hence the name). To decode:

1. Look up each character in the Base64 alphabet to get a 6-bit value.
2. Concatenate the 6-bit chunks back into a bit stream.
3. Split the bit stream into 8-bit bytes.
4. The = padding characters are dropped — they only exist to mark how much of the last group is real.

Example: SGVsbG8= → 4 chars = 24 bits = 3 bytes = 0x48 0x65 0x6c plus the last group of one character which represents 1 byte = 0x6f. Reassembled:0x48 0x65 0x6c 0x6c 0x6f= "Hello".

Decoding UTF-8 Text Correctly

The single most common Base64 decoding bug is UTF-8 mangling. Here is what happens and how to fix it:

// Encoded: 'Hello 世界 👋' as UTF-8, then Base64
const encoded = 'SGVsbG8g5LiW55WMIPCfkYs=';

// WRONG — atob() returns a "binary string" of raw byte codes
atob(encoded);
// → "Hello 世界 👋"  (garbled!)

// CORRECT — decode the bytes as UTF-8
new TextDecoder().decode(
  Uint8Array.from(atob(encoded), c => c.charCodeAt(0))
);
// → "Hello 世界 👋"  (correct)

Handling URL-Safe Base64 and Missing Padding

function base64UrlDecode(str) {
  // 1. Restore standard Base64 alphabet
  let b64 = str.replace(/-/g, '+').replace(/_/g, '/');
  // 2. Restore padding
  const pad = b64.length % 4;
  if (pad) b64 += '='.repeat(4 - pad);
  // 3. Decode + UTF-8 render
  return new TextDecoder().decode(
    Uint8Array.from(atob(b64), c => c.charCodeAt(0))
  );
}

// Works for JWT payloads
const jwt = 'eyJ1c2VyIjoi5LiW55WMIn0';  // no padding
console.log(base64UrlDecode(jwt));      // → {"user":"世界"}

Decoding a Base64 Image

// Common case: a data URL like "data:image/png;base64,iVBORw0KG..."
function decodeBase64Image(dataUrl) {
  const [prefix, b64] = dataUrl.split(',');
  const mimeMatch = prefix.match(/data:([^;]+);base64/);
  const mime = mimeMatch ? mimeMatch[1] : 'application/octet-stream';

  const bytes = Uint8Array.from(atob(b64), c => c.charCodeAt(0));
  const blob = new Blob([bytes], { type: mime });
  return URL.createObjectURL(blob);
}

// Use it
const imgSrc = decodeBase64Image(myDataUrl);
document.getElementById('preview').src = imgSrc;

Common Decoding Errors

• InvalidCharacterError: Failed to execute 'atob' — the input contains characters outside the Base64 alphabet (whitespace, quotes, non-ASCII). Strip whitespace with str.replace(/\s+/g, '') before decoding.
• Output is garbled UTF-8 — you used atob() directly. Wrap it in TextDecoder as shown above.
• Length is not a multiple of 4 — padding is missing. Add it manually:str + '='.repeat((4 - str.length % 4) % 4).
• Result looks like binary garbage — the source was probably a file (image, PDF, etc.), not text. Treat the output as bytes, not a string.
• Silent truncation — some copy-paste operations drop trailing= characters. Verify your input string length before decoding.

Key Facts

Native function:

window.atob() in every browser since IE10

Output type:

Binary string (one char = one byte) — needs TextDecoder for UTF-8

Size ratio:

Base64 → bytes is 4:3 (decoded is ~75% of encoded)

Character set:

A-Z, a-z, 0-9, +, /, = padding (URL-safe uses - and _)

Common uses:

JWT payloads, data URLs, HTTP Basic Auth, MIME attachments

Server round-trip:

None — all decoding happens in your browser

Related Base64 Tools

• Decode Base64 String — string-focused decoder walkthrough
• Base64 Decode JavaScript — atob() and TextDecoder patterns
• Base64 Decode Python — base64.b64decode() guide
• Base64 Image Decoder — decode data URLs to viewable images
• Base64 Encode Online — reverse direction
• JWT Debugger — decode entire JWT tokens