CSV to JSON

updated 12 March 2026

Before you start

You need one of the following:

• A CSV file on your computer (.csv, .tsv, or plain .txt — any delimiter is auto-detected), or
• CSV text you can paste directly into the input pane.

The first row is treated as the header row and its values become the keys in the output JSON. If your file has no header, add one before pasting, or you'll end up with keys like "42", "Alice" taken from the first data row.

There is no file size cap, but your browser's memory is. As a rough rule of thumb, files under 200 MB convert without drama on a modern laptop; above that, consider the JSON Lines option so nothing has to fit into a single string.

How to use it

1. Paste your CSV into the left pane, or drop a .csv file anywhere on the pane.
2. Pick a Format: JSON array for a single document, or JSON Lines for one object per line.
3. Toggle pretty-print if you want indentation (handy for reading, terrible for file size).
4. Review the output on the right — the status bar at the bottom tells you how many rows were parsed.
5. Click Copy to put the result on your clipboard, or Download .json to save a file.

Options explained

JSON array vs. JSON Lines

JSON array gives you a single valid JSON document: [{...}, {...}]. Use this when the consumer expects one blob — e.g. pasting into a REST client, feeding a frontend that does JSON.parse(), or storing a small reference file.

JSON Lines (also called NDJSON) is one JSON object per line, with no outer brackets and no commas between records. Use this for big files, log pipelines, BigQuery / Snowflake / S3 imports, and anything that streams record-by-record. It's also much faster to append to.

Pretty-print

Off produces a single compact line per record. On adds two-space indentation for arrays and readable newlines inside each object. For JSON Lines, pretty-print is ignored: the format requires one record per line.

Example

Input:

id,name,active
1,Alice,true
2,"Bob, Jr.",false

Output (JSON array, pretty-printed):

[
  { "id": "1", "name": "Alice", "active": "true" },
  { "id": "2", "name": "Bob, Jr.", "active": "false" }
]

Notice that id and active come out as strings, not numbers or booleans. That's deliberate — see the FAQ below.

Tips & common pitfalls

• Values are strings, by design. A cell like 007 stays "007". If this tool auto-converted, you'd lose leading zeros, scientific-notation IDs, and "false" vs. the literal word. Cast to numbers/booleans downstream where you control the logic.
• UTF-8 BOM is handled. Excel likes to prepend  to exported CSVs; the parser strips it so the first key isn't "id".
• Delimiter auto-detection samples the first few lines. If your file is mostly numeric and the first few lines have the "wrong" structure, the parser may pick semicolons by mistake — add a clearer header row if you see odd output.
• Embedded commas and newlines are fine if the cell is double-quoted, per RFC 4180. "Bob, Jr." parses as one cell.
• Duplicate header names will overwrite each other in the JSON object. Rename duplicates in the source before converting.

Troubleshooting

Frequently asked questions