CSV Splitter: Split a Large CSV into Smaller Files

How to split a CSV file

1. 1Choose your file. Drop it onto the splitter at the top of this page, or click “Choose a file”. CSV, TSV and TXT files of any size work. No file to hand? “Try it with sample data” loads a year of made-up UK weather.
2. 2Choose how to split it. By rows (1,000,000 rows per part by default, so every part opens in Excel), by size in MB, or by a column, which makes one file per value, such as one per city.
3. 3Split and download. Click “Split & download”. The parts come as one ZIP file; in Chrome and Edge on a computer you can save them straight into a folder instead. The list shows every part with its rows and size.

Split by rows, by size or by a column

By rows: parts that open in Excel

Every part gets the same number of rows under the header, and the last one what is left. The default, 1,000,000 rows, opens in Excel, which stops at 1,048,576 rows. Blank lines stay where they are, but they don’t count as rows.

By size: parts under a limit

Each part is at most the size you choose, the header included: handy for email attachments and upload limits. Parts end between two rows, never in the middle of one. A single row bigger than the limit gets a part of its own, and the tool tells you.

By a column: one file per value

Pick a column such as a city, a product or a month, and every value gets its own file, named after it: weather-London.csv, weather-Leeds.csv. The rows keep their order within each file. Up to 1,000 different values: a column with more, such as a timestamp, is refused with a message instead of making thousands of files.

Your rows, exactly as they were

The splitter never rewrites a row. Numbers, dates, leading zeros, quotes and line endings stay exactly as they are in your file, and a quoted value with a comma or a line break in it stays whole, because a part only ends at the end of a row. Lines above the header, such as a logger’s notes, and a units row under it are repeated with the header. Split by rows or by size, the parts put back together without their repeated headers are your original file, byte for byte.

A worked example: a year of UK weather

The sample behind “Try it with sample data” is a year of made-up hourly weather for six UK cities (London, Manchester, Leeds, Cardiff, Edinburgh, Belfast): 52,560 rows, 2.16 MB, one row per city per hour. Its first lines:

time,city,temperature,humidity,sky
2025-01-01 00:00,London,3.1,90,Cloudy
2025-01-01 00:00,Manchester,3.5,89,"Rain, light"
2025-01-01 00:00,Leeds,1.7,91,Showers
2025-01-01 00:00,Cardiff,2.6,98,"Rain, light"
2025-01-01 00:00,Edinburgh,2.2,92,Cloudy
2025-01-01 00:00,Belfast,2.3,91,Clear

Split by rows, 10,000 per part, it makes 6 files, each with the header on top:

Split by the city column instead, it makes 6 files of 8,760 rows each, one per city, the rows still in time order:

The sky column holds values such as "Rain, light", with a comma inside the quotes: they stay one value in every part.

Why split a CSV for Excel?

Excel opens at most 1,048,576 rows per sheet, and a longer CSV is cut off at that row. Splitting it into parts of 1,000,000 rows gives you files that each open whole. Our guide CSV too large for Excel? explains the limit and the other ways around it.

Don’t need Excel at all? The large CSV viewer scrolls, sorts and filters every row of the whole file. To know how many parts you will get, the CSV row counter counts the lines of any file in seconds. And to put parts back together, or combine exports, use Merge CSV files.

Private by design: nothing is uploaded

Online CSV splitters that upload your file send a copy to their server. MegaRows does not: the splitting runs entirely in your browser, on your device, and neither the file nor its name is sent anywhere. Read the privacy statement for details.

Other ways to split a CSV

macOS and Linux: split

split -l 1000000 big.csv part-

Fast and built in, but the header ends up in the first part only, and a quoted value with a line break in it can be cut in two between parts.

Python: pandas

import pandas as pd
for i, part in enumerate(pd.read_csv("big.csv", chunksize=1_000_000)):
    part.to_csv(f"part-{i + 1:03}.csv", index=False)

Repeats the header in every part, but every value is read and written again: numbers can change their format and leading zeros disappear unless you read every column as text with dtype=str.