Split a CSV into smaller files

How to split a CSV file

1. Open a .csv, .tsv or .txt file, drop it on the tool, or paste rows. The first row is treated as the header.
2. Choose what to split by and enter the number of rows, the maximum size in bytes, the number of files, or the column.
3. Preview the parts to see each file's name, row count and size, then download a single part or all of them as one ZIP.

Parts are named after the original file: orders.csv becomes orders-part-001.csv, orders-part-002.csv… A .tsv file keeps the .tsv extension. The ZIP is named orders-parts.zip.

Ways to split

Quoted values, raw records and file names

• Quoted values containing the delimiter or line breaks stay inside their record, read per RFC 4180. A multiline address is never split across two parts.
• Records are copied as raw bytes, never re-parsed and re-written: quotes, line endings and leading zeros are kept exactly. A UTF-8 byte-order mark at the start of the file is written at the start of every part, so Excel still recognizes each one as UTF-8.
• Blank lines between records are skipped. If the last record has no line break, the part gets the file's own line ending.
• Column values are compared exactly: West and west become separate files. Since many file systems ignore case, the second one is named orders-west-2.csv rather than overwriting the first.
• In file names, characters such as / \ : * ? " < > | become _, names are cut at 60 characters, and an empty value is written as orders-empty.csv. Use the value column in the preview to see exactly which value each file holds.

Encodings that can be split

Splitting works directly on bytes, which is safe for UTF-8 and single-byte encodings such as Windows-1252. UTF-16 files cannot be split this way; save them as UTF-8 first, for example with Change CSV delimiter and encoding. Shift_JIS, GB18030, GBK and Big5 files can be split when the delimiter is a comma, semicolon or tab, but not a pipe, because a pipe byte can occur inside a character in those encodings. The tool tells you before you start when a file can't be split.

Split errors and how to fix them

• One record is larger than the chosen part size. Raise the maximum size; it has to fit the header plus the largest single record.
• This split would create more than 1,000 files. Increase the rows per file or the size, or choose fewer files.
• This column has more than 1,000 distinct values. Split by a broader column, such as region instead of customer, or filter the file first.
• A quoted value never closes. A quote opened somewhere in the file is never closed, so the rest of the file would be one record. Find it with Validate CSV.

Large files and privacy

Split is not limited to 25 MB. A larger file shows a preview of its first 200 rows, and the split reads the whole file in 8 MB chunks, so memory use stays low. Parts are kept as browser file data until you download them; the practical limit is the free storage your browser can use.

Splitting runs in a background worker in this tab and Cancel stops it. The file is never sent to a server. To put parts back together, use Merge CSV files; to shrink a file instead of cutting it up, use Compress CSV.

Example: an order export split into parts of 2 rows

A small order export with CRLF line endings (shown here as plain line breaks) and a multiline note, split into parts of 2 rows with the header repeated. The quoted line break stays inside order A2, and the last record, which had no line break in the input, gets the file's CRLF ending.

order_id,region,note
A1,West,ok
A2,East,"two
lines"
A3,West,"a, b"
A4,,late
A5,east,ok

orders-part-001.csv (2 rows)
order_id,region,note
A1,West,ok
A2,East,"two
lines"

orders-part-002.csv (2 rows)
order_id,region,note
A3,West,"a, b"
A4,,late

orders-part-003.csv (1 row)
order_id,region,note
A5,east,ok