How to use this tool
- Load a CSV with the correct delimiter and header setting.
- Choose data records per part and whether to repeat the header.
- Split and download each part; the manifest lists each file’s record count.
How the operation works
Splitting happens after parsing, so quoted multiline fields stay together. The final file contains the remaining records, even when that is fewer than the selected batch size. Part names are sequential. The 100-part limit prevents accidental generation of thousands of downloads.
Split by records, not physical lines
Splitting happens after parsing, so quoted multiline fields stay together in one record. You choose data records per part and whether to repeat the header in each file. Repeated headers do not count toward the batch size. The final part contains remaining records even when fewer than the selected batch size.
Part names are sequential (part-001.csv, part-002.csv, …). A 100-part limit prevents accidental generation of thousands of downloads.
Example split
Three data records; two records per part part-001.csv: header + 2 data records part-002.csv: header + 1 data record
The main download is a manifest describing each part and its record count. Use each numbered Download button to save its CSV file. Verify that the counts add up to the original before submitting to an importer.
Row count is not byte size
A row-count limit is not a byte-size limit. Records vary greatly in length. If an import has a strict megabyte cap, inspect each part’s file size after download—not only the record counts. The 10 MiB input and 50,000 record limits still apply to the source.
When to split: importer rejects large uploads; you need per-batch review; or a downstream process expects fixed-size chunks. When not to: when the destination accepts the full file—unnecessary splits add reassembly work.
Split safely
1) Inspect in CSV Viewer. 2) Align schema if needed (Column Editor). 3) Split with header repeat on for standalone parts. 4) Download parts and the manifest. 5) Spot-check one part in the destination. Keep the original intact.
For combining aligned files later, see CSV Merge.
A worked example
Three data records; two records per part
part-001.csv contains two data records; part-002.csv contains one. Repeated headers do not count toward the batch size.
A row-count limit is not a byte-size limit. Records can vary greatly in length. The main download is a manifest describing the parts; use each numbered Download button to save its CSV file. Inspect those file sizes if an import has a strict megabyte cap, and verify that their data-record counts add up to the original.
Splitting records, not physical lines
Split counts data records, not raw newlines. A quoted field can contain line breaks and still belong to one record. The header is repeated in every part when that option is enabled; otherwise only the first part carries it. Each part is generated only when you download it, and record order is preserved.
The part size is a number of data records, not bytes. A part containing very wide rows can still exceed an upload limit even when the record count looks small. Test one part in the destination before splitting a large export.
Choosing a part size
Start from the destination limit and the average row width. If the limit is in bytes, estimate from a sample and leave headroom for the header and quoting. If the limit is a row count, use that number directly. Keep the original file and a note of the split settings so the parts can be regenerated.
Reassembling parts requires care: repeated headers must be removed and the order restored. Prefer keeping the original file for archival use rather than joining the parts back together.
Processing and privacy
CSV Split divides the table locally and creates parts only when you download them; the source file is not uploaded. Input stays in browser memory until you reset or leave the page. Copy and download run only when you choose them. No tool input is sent to a processing service. Website pages and scripts still make ordinary hosting requests; the privacy policy separates those from tool processing.
Frequently asked questions
Will a field with a line break be split?
No. A quoted newline remains inside its original CSV record.
Do repeated headers count toward the batch size?
No. Batch size counts data records only.
What does the main download contain?
A manifest listing each part file and its record count. Individual CSVs are separate downloads.
How many parts can I create?
Up to 100. The source is also limited to 10 MiB and 50,000 records.
Can I split by file size instead of records?
Not in this release. Split by records, then check each part’s bytes if a size cap matters.
Are parts split by file size?
No. You choose a number of data records per part; byte size can vary with row width.
Does every part include the header?
Only when the repeat-header option is enabled.
How this tool was checked
CSV Split was checked with quoted newlines, repeated headers, one-record parts, exact multiples of the part size and files near the 50,000-record limit. See the testing method or report a failure.