Unexpected End of JSON Input

Quick answer: Unexpected end of JSON input means JSON.parse() reached the end of the string before the value was complete — the data was empty or cut off. It's different from "unexpected token": there the parser found a wrong character; here it found no more characters. The fix is to make sure you're parsing a complete, non-empty body. To confirm where a document stops short, paste it into the JSON Debugger, which shows the last valid position.

JSON.parse("") throws this error, and so does JSON.parse('{"a":1') — an object that was opened but never closed. The parser consumed everything you gave it and was still waiting for more (a closing }, a closing quote, the rest of a number). Reaching the end of input in that state is the error.

JSON.parse("");          // Unexpected end of JSON input
JSON.parse("{");         // Unexpected end of JSON input
JSON.parse('{"a":1,');   // Unexpected end of JSON input

• Empty response body. An API returned 204 No Content, or a DELETE/POST succeeded with no body, and you still called .json() on it.
• Truncated response. A network timeout, a proxy buffer limit, or a body-size cap cut the JSON off mid-stream, so the closing brackets never arrived.
• Unclosed structure in hand-written JSON — a missing }, ], or closing ".
• Reading a file that hasn't finished writing, or a stream you parsed before it completed.

Guard against empty bodies before parsing:

const res = await fetch("/api/save", { method: "POST" });
const text = await res.text();
const data = text ? JSON.parse(text) : null;   // 204/empty → null, not a throw

For truncation, the JSON itself is incomplete — re-fetch it rather than trying to parse a half-document. Check the response Content-Length against what you received, and look for timeouts or proxy limits in the middle of large payloads.

For hand-written JSON, the JSON Debugger jumps to the last position it could parse, which is usually right where a bracket or quote is missing. Then run the result through the JSON Validator to confirm it's complete.

When an LLM or a logging system cut a document off and you genuinely need to recover the partial data, an automatic repair pass can close the open brackets and strings. See Repairing Truncated LLM JSON for that workflow, or clean model output with the AI JSON Cleaner. For a complete document, prevention is better: always serialize with JSON.stringify() rather than concatenating strings — see Parse and Stringify.

What causes "Unexpected end of JSON input"? The string passed to JSON.parse() was empty or incomplete. Common sources are a 204 No Content response, a truncated network reply, or JSON with a missing closing bracket or quote.

How is it different from "Unexpected token"? "Unexpected token" means the parser found an invalid character somewhere in the middle. "Unexpected end of JSON input" means it ran out of characters before the value finished — nothing was wrong, there just wasn't enough.

How do I stop empty API responses from throwing? Read the body as text first and only parse it if it's non-empty: text ? JSON.parse(text) : null. This handles 204 and empty success responses cleanly.

Can I recover truncated JSON? Sometimes. If the cut-off is near the end you can auto-close the open brackets to salvage most of the data, but a re-fetch of the complete response is more reliable when possible.