Base64 Encode & Decode

What Base64 actually is

Base64 is an encoding — a way to represent arbitrary binary data using only the 64 printable ASCII characters A–Z, a–z, 0–9, +, /. It is defined by RFC 4648 and was invented in the 1980s so binary attachments could survive email systems that only handled 7-bit text.

It is not encryption. Anyone with the encoded string can decode it back to the original — there is no key, no secret. If you need confidentiality, use AES or your TLS layer; Base64 only solves "how do I transport bytes through a text-only channel?".

The transform itself is mechanical: every 3 bytes of input become 4 output characters. Input that doesn't divide evenly into 3-byte groups gets padded with = so the output length is always a multiple of 4. This is why you see = at the end of some encoded strings (zero, one, or two of them).

When you'll reach for it

• Data URIs. Inlining a small image into CSS or HTML: url("data:image/png;base64,iVBOR…").
• JWT. Each segment of a JSON Web Token is Base64URL-encoded JSON.
• HTTP Basic auth. The Authorization: Basic … header is user:pass Base64-encoded.
• Email attachments (MIME). Binary files survive transit by being Base64-wrapped.
• API payloads. Sending binary blobs over JSON (which has no native binary type).
• Storing binary in databases or environment variables that expect text.

You should not use Base64 to "shrink" data — it adds ~33% overhead. It's a transport encoding, not a compression.

The UTF-8 gotcha

The browser's built-in btoa() function looks like it does the right thing — and silently produces garbage the moment you feed it a non-ASCII character:

btoa('Hello')      // "SGVsbG8="              ✓
btoa('世界')        // InvalidCharacterError   ✗
btoa('🌏')         // InvalidCharacterError   ✗

The reason: btoa operates on Latin-1 (bytes 0–255), not UTF-8. The character 世 doesn't fit in one byte, so the function refuses. Same for emoji, Vietnamese with diacritics, Arabic, anything outside basic Latin.

The fix is to UTF-8 encode the string first into a byte sequence, then Base64-encode those bytes. That's exactly what this tool does — every input is treated as UTF-8 by default, so emoji and unicode round-trip cleanly.

Base64 vs Base64URL

Standard Base64 uses + and /, which both have special meanings in URLs and filenames. Base64URL (RFC 4648 §5) swaps them for - and _, and strips the trailing = padding. This makes encoded strings safe to drop into a URL query parameter, a JWT segment, or a filename.

If you toggle URL-safe above, the tool emits - and _ instead of + and /, with no padding. Decoding accepts either flavor — we auto-detect.

Examples

Encode plain text:

Input: Hello, World! Output: SGVsbG8sIFdvcmxkIQ==

Encode Vietnamese with diacritics:

Input: Tiếng Việt có dấu Output: VGnhur9uZyBWaeG7h3QgY8OzIGThuqV1

Decode a JWT payload chunk (Base64URL):

Input: eyJzdWIiOiIxMjM0NTY3ODkwIn0 Output: {"sub":"1234567890"}

Security notes

• Never store passwords as Base64. It's reversible. Use a password hash function (bcrypt, argon2, scrypt) instead.
• Don't trust Base64-decoded JSON without re-validating. A JWT signature is what proves integrity; the Base64 layer alone proves nothing.
• Watch input size when decoding. A few KB of Base64 expands to a few KB of bytes — fine — but a few hundred MB at once can crash a browser tab. This tool processes input synchronously; for very large files prefer a streaming approach.

Privacy

This page runs the encoding/decoding logic entirely in your browser using the open-source js-base64 library. Your input is never sent to our servers, never logged, and never leaves your device. Open the network tab and try it — you'll see zero requests.

Related tools

• JWT Decoder — inspect a JSON Web Token (which is three Base64URL chunks).
• URL Encoder — for percent-encoding in URLs, a different (often complementary) encoding.

Is Base64 encryption?

No. Base64 is a reversible encoding — anyone holding the encoded string can decode it back to the original without a key. Use it to transport binary data through text-only channels, not to keep secrets.

Why does btoa() break on emoji or Vietnamese text?

The browser's native btoa() only handles Latin-1 (bytes 0–255). Characters like 世 or 🌏 are multi-byte in UTF-8 and don't fit, so btoa throws. This tool encodes UTF-8 first, then applies Base64 — so unicode survives the round trip.

What's the difference between standard Base64 and Base64URL?

Standard Base64 uses + and / and pads with =. Base64URL replaces +/ with -_ and drops the padding, making the result safe to use in URLs, JWT segments, or filenames. Toggle "URL-safe" above to switch alphabets.

Why do I see = at the end of some encoded strings?

Base64 encodes every 3 input bytes as 4 output characters. When the input length isn't a multiple of 3, the encoder pads the output with one or two = characters so the total length stays divisible by 4. Base64URL omits this padding.

Is my input sent to a server?

No. All processing happens in your browser via the js-base64 library. Open your browser's network tab — you will see zero requests while you type. Once the page has loaded you can disconnect entirely and the tool keeps working.

What's the largest input this can handle?

Practically, a few megabytes per operation. Beyond ~10 MB you may notice the browser tab pausing while the conversion runs synchronously. For multi-hundred-MB files, prefer a streaming decoder in a Web Worker or a server-side tool.

My decoded text looks like garbage — what went wrong?

Almost always one of three things: (1) the input is encoded with a different alphabet (URL-safe vs standard — try toggling), (2) the input was truncated and is missing padding, or (3) the original wasn't UTF-8 text but binary that happens to be invalid UTF-8 (then "decoded text" is meaningless; you want the raw bytes).

Can I use Base64 to compress data?

No — Base64 makes data ~33% larger, not smaller. If you need to shrink, use gzip, brotli, or a domain-specific codec, then Base64-encode the compressed bytes if you need to transport them as text.