Base64 to Text Decoder Online — Free

Something outside the Base64 alphabet slipped into the input. Usually it is quotes copied along with a JSON value, soft line breaks from a mail client, or the truncating ellipsis from a console dump. Strip everything except letters, digits, plus, slash and the trailing equals signs, then decode again.

Padding. Base64 processes input three bytes at a time, and when the final group holds only one or two bytes the output is padded with equals signs to a multiple of four characters. One sign means a single leftover byte, two means a pair. It carries no data of its own.

Base64 stores bytes and knows nothing about character encodings. Text originally encoded as UTF-8 decodes cleanly, but a string produced from a legacy single-byte encoding falls apart: the bytes are right, they are simply being read as UTF-8. That case needs a separate re-encoding step after decoding.

Tokens use the base64url variant: hyphen instead of plus, underscore instead of slash, and the padding usually dropped. Swap the hyphens and underscores back and pad the length to a multiple of four. For a full token breakdown the JWT Decoder page is the better place.

No — the string is decoded by a script in your own tab and is never transmitted, and browser history does not record field contents. It does stay visible on screen until you press clear, so wipe the boxes before sharing your screen with anyone.

You will see a jumble of control characters, which is what a PNG or PDF looks like when read as text. For binary payloads use the Base64 to File page, which rebuilds a downloadable file with the extension you choose, or Base64 to PDF when the payload is a document.