Base64 Decode

What the decoder accepts

Decoding reverses the encoding: each character is mapped back to its 6-bit value and every four characters become three bytes. Base64 found in the wild is rarely clean, so this decoder is forgiving:

• Spaces and line breaks are ignored, so wrapped MIME bodies and PEM blocks can be pasted as they are. Delete the -----BEGIN and -----END lines first, though: their letters and dashes are valid Base64 characters and would be decoded as data.
• URL-safe input with - and _ decodes just like standard Base64 with + and /.
• Missing padding is fine. Only a length of 4n + 1 characters is rejected, because a single leftover character cannot form a byte; it usually means the string was cut off.
• Data URIs work too: a leading data:...;base64, prefix is removed before decoding.

Text or binary output

The result of decoding is always a sequence of bytes. If the original was text, keep text output and pick its character set. If it was an image, PDF, archive, key or ciphertext, no character set will make it readable: switch the output to hex to inspect the bytes, or open Base64 to File or Base64 to Image to preview and save it. The first characters often give the content away:

Why the decoded text looks garbled

• Wrong character set. UTF-8 text read as Windows-1252 turns é into é, and GBK or Shift_JIS text read as UTF-8 shows replacement characters (�). Select the charset the text was written in.
• Binary data. Random symbols with the odd readable fragment, such as PNG or PK near the start, mean the content is a file rather than text.
• Compressed or encrypted data. gzip streams, Protocol Buffers and ciphertext decode correctly but are not text. Hex output shows what is really there.
• Double encoding. If the output is itself Base64, paste it back into the input and decode it a second time.

Decoding Base64 in code

# Python
base64.b64decode(s).decode('utf-8')

// Node.js (also accepts URL-safe input)
Buffer.from(s, 'base64').toString('utf8')

// Browser: atob() returns one character per byte, so decode UTF-8 afterwards
new TextDecoder().decode(Uint8Array.from(atob(s), c => c.charCodeAt(0)))

# Linux and macOS shell
echo 'SGVsbG8=' | base64 --decode