DoItAl

Data utility

CSV Split & Partition

Split a CSV into smaller files by row count, number of files, or column value—without uploading your data.

Browser-local • No upload

1. Choose file

CSV, TSV, semicolon, and pipe-delimited text up to 10 MB.

Drop a CSV or TSV file here

or choose a file from this device

2. Choose how to split

3. Configure

4. Review

Review the complete plan before creating any downloads.

Choose a file to build the split plan.

5. Download

Every output is generated locally. Individual downloads are also available in the plan above.

Need to clean duplicates, blank rows, or messy headers first? Use CSV Cleaner. Need the opposite job? Combine files with CSV Merge & Join.

Continue with

How to split a CSV

Choose a local CSV or TSV, select a split method, review the exact output plan, then download every part in one ZIP. No file contents leave your browser.

Split by rows vs column value

Use rows per file when an import system has a row or file-size limit. Use column value when stores, teams, clients, regions, or categories each need a separate file. When splitting into a fixed number of files, earlier files receive one extra row until the remainder is used: 10 rows across 3 files becomes 4, 3, and 3.

Are headers preserved?

Yes. Repeating the source header in every output is on by default so every part stays independently usable. Turn it off when the receiving system requires data rows only. If the input has no header, none is invented.

Is my CSV uploaded?

No. Parsing, planning, splitting, serialization, and ZIP creation all happen in this browser tab. Analytics never include filenames, headers, column names, values, generated names, or file contents.

Why can a column create too many files?

A nearly unique column—such as email or order ID—can produce one file per row. V1 blocks plans above 500 output files, the tested browser-local limit, and recommends splitting by rows or target file count instead.

Private, bounded processing

V1 accepts one file up to 10 MB, keeps ZIP work below a 40 MB uncompressed-output guardrail, and previews only the first 50 input rows and 50 planned outputs. It is intentionally memory-based rather than a multi-gigabyte streaming engine.

What this tool does

Partitions a CSV/TSV by rows per file, target file count, or exact column value and downloads the planned parts in a ZIP.

How to use it

  1. Load a table and confirm header behavior.
  2. Choose a split method and review the output plan.
  3. Download the ZIP and verify a part against the receiving system’s requirements.

Worked example

Ten data rows across three files become 4, 3, and 3 rows. With repeated headers enabled, each file also includes the source header.

Processing and explanation

Count splitting distributes remainder rows to earlier parts. Column grouping uses exact values; a high-cardinality identifier may create almost one file per row.

Limits and common mistakes

One input up to 10 MB; maximum 500 output files and a 40 MB uncompressed-output guardrail. Preview shows at most 50 input rows and 50 planned outputs. It is not a multi-gigabyte streaming splitter.

Privacy and your input

Source data, group values, generated filenames, and ZIP contents stay on this device.

Frequently asked questions

Does it invent headers?
No. Header repetition uses the source header when the input has one.
Can I select records by a condition?
Use CSV Filter & Extract Rows for conditions; this tool creates partitions.

Material content update: