CSV File Size Limits (Excel, Sheets, Databases)

"File too large." "Not all data loaded." "Application not responding."

A CSV has no built-in row limit. The practical limit comes from the application, available memory, and what you need to do with the file.

Technically, CSV files have no size limit. A CSV is just text. It can be 1KB or 1TB. The limit comes from the software you use to open it.

Here are the hard limits for the most common tools.

1. Microsoft Excel

The Limit: 1,048,576 rows. The Behavior: If you open a CSV with 2 million rows, Excel will load the first 1,048,576 and simply delete the rest. It usually gives a warning: "Text file contains more data than will fit on one worksheet."

Column Limit: 16,384 columns (Column XFD).

Workaround:

• Split the CSV into multiple files.
• Use Excel's "Power Query" (Data > Get Data) to link to the file without loading it all into the grid.

2. Google Sheets

The Limit: 10 million cells. The Behavior: This is a cell limit, not just rows.

• If you have 1 column, you can have 10 million rows.
• If you have 10 columns, you can have 1 million rows.
• If you have 100 columns, you can have 100,000 rows.

File Size Limit: 100 MB for import.

Workaround: None. You must split the data.

3. Text Editors (Notepad, VS Code)

Notepad (Windows): Struggles with files > 500MB. Notepad++: Can handle larger files (up to 2GB on 64-bit), but gets slow. VS Code: Optimized for code, not large data. Will warn you on large files. Vim / Sublime Text: Generally handle large files better than standard editors.

The Limit: Usually limited by your computer's RAM. If you have 16GB RAM, you can't open a 20GB text file in a standard editor because it tries to load it all into memory.

Solution: Use "Large File Viewer" tools (like LTFViewer) that stream data from disk.

4. Databases (MySQL, PostgreSQL)

The Limit: Effectively unlimited. The Behavior: Databases are designed for this. A table can hold billions of rows. Import Limit: The LOAD DATA command has no inherent limit, though server configurations (max_allowed_packet) might need tweaking.

Solution: This is the correct place for large data. Import your large CSV into a database, then query just the slice you need.

5. Python (Pandas)

The Limit: Your RAM. The Behavior: pd.read_csv() loads the entire file into memory. A 1GB CSV might take 2-3GB of RAM to process. If you run out of RAM, Python crashes (MemoryError).

Solution: Process in chunks.

for chunk in pd.read_csv('large_file.csv', chunksize=10000):
    process(chunk)

Summary Table

Tool	Row Limit	File Size Limit (Approx)
Excel	1,048,576	~100-200MB (practical)
Google Sheets	10M cells	100MB import
Notepad	N/A	~500MB
MySQL/Postgres	Unlimited	Unlimited
Python (Pandas)	RAM dependent	RAM dependent

What to Do with a "Too Big" CSV

1. Split It: Break it into chunks of 1 million rows. -> Split Tool
2. Filter It: Use a command line tool or script to extract only the rows you need (e.g., "Only rows from 2024").
3. Database It: Import to SQL and use queries.

Measure the file before choosing a method

File size alone is a poor predictor of memory use. A narrow 500 MB file of simple numbers may be easier to process than a smaller file containing very long quoted text. Parsers also create in-memory objects that can require several times the bytes stored on disk.

Start by recording the byte size, approximate record count, column count and longest fields. Decide whether you need every column and row. Filtering at the source is usually faster than downloading everything and discarding most of it later.

For a repeatable pipeline, process in chunks and write each result incrementally. Avoid accumulating every chunk in one list, which defeats the purpose of chunking. Reconcile the processed record count against the source and log rejected records separately. If the work needs joins, repeated queries or concurrent access, load the data into a database or a columnar format instead of repeatedly scanning CSV.

---

File too big for Excel? HappyCSV's Split Tool can chop your large CSV into Excel-safe chunks without installing software.