← Guides

Developer Guide

Base64 vs URL Encoding

Base64 turns bytes into ASCII. Percent-encoding makes a value legal in a URL. They look similar and solve different jobs — with examples from the browser tools.

base64 vs url encodingpercent encoding vs base64encodeuricomponent versus base64when to use base64when to use url encoding

Both encodings emit printable ASCII, which is why they get swapped in code reviews. Base64 is a binary-to-text alphabet: any byte sequence becomes A–Z, a–z, 0–9, +, /, and =. Percent-encoding (URL encoding) is a quoting rule for URLs: a character that would break the address is replaced with % plus two hex digits.

Use Base64 when the data is bytes in a text channel. Use URL encoding when the data is already text inside a URL. Encoding a search term as Base64 wastes space and still leaves + and / unsafe in a query.

What Base64 is actually doing

Base64 groups input in threes of bytes and writes four characters. The payload grows by about 33%. Padding = fills the last group when the length is not a multiple of three. /tools/base64 encodes Unicode text this way: hello becomes aGVsbG8=.

That string is safe in JSON, email, and a data: URI. It is not a secret. Anyone who can read aGVsbG8= can decode it. Standard Base64 is also not URL-safe: +, /, and = are reserved or awkward in queries. The tool’s URL-safe mode uses - and _ and drops padding — only when the receiver expects Base64url (JWTs, some query tokens).

What percent-encoding is actually doing

A URL has syntax (?, &, =, #) and data (the search box). Percent-encoding protects the data. A space becomes %20. hello world on /tools/url-encoder in encodeURIComponent mode is hello%20world.

encodeURIComponent is for one value. encodeURI leaves : / ? # and friends alone so a whole address stays structured — which means it will not encode an & inside a parameter. HTML forms may turn spaces into + instead of %20. If a plus sign is data (C++), encode it as %2B before a form decoder sees it.

Size, alphabet, and reversibility

Base64 always expands. Percent-encoding only expands the characters that need it; hello is unchanged in a query, while hello world picks up three extra characters for the space. Both mappings are reversible. Neither is a hash: the original bytes come back exactly.

The alphabets overlap on letters and digits and then diverge. If you see a long block of mixed case with trailing =, you are looking at Base64. If you see %20, %26, or %2F, you are looking at a URL.

When you need both

A data URI is Base64 inside a URL: data:image/png;base64,iVBOR…. Browsers parse that as a data URL, so the + and / in the payload are fine after the comma. The same payload as ?img= is a query value: percent-encode it, or switch to Base64url first.

The correct stack is Base64 (or Base64url) first, then encodeURIComponent if the string becomes a query parameter. Encoding twice with encodeURIComponent turns %20 into %2520 and is a bug, not a second layer of safety.

A practical rule

Bytes, files, and JSON fields that must stay text → /tools/base64. A string going into a query, path, or form field → /tools/url-encoder with encodeURIComponent on the value, not the whole URL.

If you are unsure, encode the same sample in both tools. One result looks like aGVsbG8=. The other looks like hello%20world. Pick the job that matches.

Common pitfalls

Dropping raw Base64 into a query is the usual breakage: + becomes a space in form decoders and / can be read as a path separator depending on the stack. Using encodeURI on a concatenated ?q= + userInput leaves & intact, so the user injects extra parameters.

Base64 of a URL is not percent-encoding. Percent-encoding of a file is not a data URI. Decode on the same tool you used to inspect the alphabet.

  • hello → aGVsbG8= on /tools/base64.
  • hello world → hello%20world on /tools/url-encoder.
  • URL-safe Base64 is a different alphabet from percent-encoding.

Worked examples

Same word, two encodings

Input
hello
Output
Base64: aGVsbG8=  ·  percent-encoding: hello

Percent-encoding leaves unreserved characters alone. Base64 always rewrites the bytes. Encode both on /tools/base64 and /tools/url-encoder.

A space in a query value

Input
hello world
Output
hello%20world

encodeURIComponent on /tools/url-encoder. Do not Base64 this search term; aGVsbG8gd29ybGQ= is a different string and still needs quoting if it later sits in a URL.

Frequently asked questions

Can I percent-encode instead of Base64 for a JSON file field?

JSON already allows Unicode strings. You need Base64 when the field is binary, not when it is text. Percent-encoding is for URLs, not JSON objects.

Why did my Base64 break in a query string?

+, /, and = are awkward in URLs. Use URL-safe mode on /tools/base64, or run encodeURIComponent on the standard Base64 string.

Is encodeURI good enough for query values?

No. encodeURI keeps & and =. Encode each value with encodeURIComponent on /tools/url-encoder, then join with &.

Does either encoding encrypt the data?

No. Both are reversible without a key. Transport secrecy is TLS.

Do these tools upload the string?

No. /tools/base64 and /tools/url-encoder run in the browser. There is no encoding API on thedevtools.in.

Summary

Base64 is for bytes in a text document. Percent-encoding is for values in a URL. Confirm hello → aGVsbG8= on /tools/base64 and hello world → hello%20world on /tools/url-encoder, and only stack them when a Base64 blob has to travel as a query parameter.

Try it yourself

Put what you've learned into practice with our free browser-native tool.

Open Tool →