JSON Errors Explained: What Each Message Means and How to Fix It

JavaScript messages also include a position, like at position 7 (line 1 column 8). Python adds line 1 column 8 (char 7). Newer Python versions word a few of these differently, but the causes are the same.

The single most common JSON error, because JavaScript, Python and most config formats allow it and JSON doesn't:

{
  "name": "Ada",
  "role": "admin",   ← this comma is the problem
}

Notice that the error points at the closing }, not the comma. The parser saw a comma, expected another property, and found the end instead. Whenever an error points at a closing bracket, look one character back.

JSON strings and keys must use double quotes. {'name': 'Ada'} and {name: "Ada"} are valid JavaScript objects but invalid JSON.

This usually happens when someone copies an object from code or prints a Python dictionary with print(data) or str(data) — that produces Python syntax with single quotes. Use json.dumps(data) in Python, or JSON.stringify(data) in JavaScript, to get real JSON.

Standard JSON has no comments at all — neither // line nor /* block */. Some files look like JSON and allow them anyway: VS Code settings and tsconfig.json use a relaxed format often called JSONC. Copy one of those into a strict parser and it fails.

Remove the comments, or move the information into a real field such as "_note": "..." if you need to keep it.

When code calls an API and then parses the response, the most confusing errors aren't about JSON syntax — the server sent something else:

• Unexpected token '<' means the response starts with < — almost always an HTML page: a 404, a server error, or a login page because your session expired.
• Unexpected end of JSON input means the body was empty, which is normal for a 204 No Content response or a request that failed silently.
• "undefined" is not valid JSON means you passed undefined — often from reading localStorage or a property that doesn't exist.

1. 1
   Check the HTTP status code before parsing.
2. 2
   Check the Content-Type header is application/json.
3. 3
   Log the first couple of hundred characters of the raw text — you'll see at once whether it's HTML or empty.

Here's a trap that catches teams with a Python backend and a JavaScript frontend. Python's json module writes NaN and Infinity by default:

json.dumps({"score": float("nan")})   →   {"score": NaN}

That isn't valid JSON, and JSON.parse in the browser rejects it with Unexpected token 'N'. Pass allow_nan=False to json.dumps to catch it on the server, and send null instead.

When the JSON looks perfect and still fails, suspect something you can't see:

• A byte-order mark at the very start, added by some Windows editors. Save the file as “UTF-8” rather than “UTF-8 with BOM”, or in Python read it with encoding="utf-8-sig".
• Smart quotes — “ and ” instead of " — from text pasted out of a word processor or chat app.
• Real tabs or line breaks inside a string. Inside JSON strings they must be written as \t and \n.

1. 1
   Paste the text into the JSON formatter. It shows the line and column of the first problem.
2. 2
   Look at that spot and one character before it — the cause is usually just before where the parser gave up.
3. 3
   Fix it and check again; there may be a second mistake further down.
4. 4
   Once it's valid, format it with indentation so the next mistake is easier to spot.

If you have a working version and a broken one, format both with sorted keys and compare them with Text Compare to see exactly what changed.