How to Save a Pandas DataFrame to CSV Without the Index

Why pandas adds an index column

Every DataFrame has an index — the labels down the left side when you print it. For a freshly built frame it is a RangeIndex: 0 through n−1, generated for you. When pandas serialises to CSV it writes those labels as the first column, because dropping data silently would violate the same principle that keepsread_csv() able to round-trip a file. The signature says so plainly:index=True.

The consequence is that the index column arrives whether it carries meaning or not. When the labels are dates, IDs or categories, that column is the point. When they are a forgotten RangeIndex, it is noise that breaks the next import — a colleague's script reads it as a column named Unnamed: 0, or the sheet simply looks wrong.

Before and after: what index=False changes

Written with pandas defaults — header row plus index column:

,name,age,city
0,Alice,25,Delhi
1,Bob,30,Mumbai
2,Charlie,28,Chandigarh

Written with index=False:

name,age,city
Alice,25,Delhi
Bob,30,Mumbai
Charlie,28,Chandigarh

The data rows are identical. The only difference is whether the labels travelling with the frame are carried into the file — and the empty header cell that goes with them.

Try it in the converter

The tool opens with the sample above and the index turned off, so the preview and the generated code already show the file you want. Tick Add row numbers to reproduce the problem — the extra column appears in the preview and in the download — then untick it. The call updates live; copy it straight into your script.

Four fixes for an unwanted index column

Fix 1 — write the file with index=False

The fix for the export itself. Put index=False in the call that writes the file and the column never reaches it:df.to_csv("output.csv", index=False). For a pandas-to-CSV export without the index, this single argument is the whole fix — and it is the only one most code needs.

Fix 2 — remove the index from the DataFrame with reset_index(drop=True)

Use this when the index should not exist anywhere — not in the file and not in the frame your next line of code touches:

df = df.reset_index(drop=True)
df.to_csv("output.csv", index=False)

The distinction matters: index=False changes only the file;reset_index() changes the DataFrame. And callingreset_index() without drop=True does the opposite of removing the index — it pushes the old labels into a new column, so writing afterwards can hand you the data twice. drop=True is what actually discards them.

Fix 3 — clean up a CSV that already has the column

The file on disk already contains the stray first column. Read it back into the index and rewrite without it:

df = pd.read_csv("output.csv", index_col=0)   # first column back into the index
df.to_csv("output.csv", index=False)           # rewrite, this time without it

If the column was never an index to begin with — it is simply junk — drop it by name. pandas names an unnamed header cell Unnamed: 0 on the way back in, sodf.drop(columns=["Unnamed: 0"]) removes it anddf.to_csv(..., index=False) keeps it gone.

Fix 4 — keep the index, but name it

Sometimes the column is right and only the empty header cell is wrong. Name it instead of deleting it:

df.to_csv("output.csv", index_label="row_id")

index_label only applies when the index is being written, so it combines withindex=True (the default) and never with index=False. The same control sits in the converter under More options → Index column name.

When you should keep the index

If the labels are real data — timestamps, order IDs, a date-indexed series — write them. The file should carry them, and the reader should be told how to get them back:

df.to_csv("output.csv")                       # index=True keeps the labels
back = pd.read_csv("output.csv", index_col=0)  # restored on the way in

Turn index=False on that frame and the round trip silently degrades: the file has no labels, so read_csv() builds a fresh RangeIndex from 0. The numbers may look identical — the original identity of each row is gone. Keep the index when it means something; drop it when it is just a counter.

Common mistakes

• Calling reset_index() without drop=True and expecting the column to vanish. It moves the index into a column instead, and the default write can then produce two extra columns where you started with one.
• Turning the index off once and assuming it is off for good.index=False belongs in every writing call — it is not a setting on the DataFrame. Keep it in the function that exports, not in an interactive edit.
• Hand-deleting the column in Excel after each export. The file will be regenerated the same way every time; fix the call that writes it.
• Renaming Unnamed: 0 after the fact. That works once, butindex_label prevents the unnamed cell from existing in the first place.
• Assuming index=False survives the round trip. The labels are not in the file, so reading it back cannot restore them — write them when you need them later.

Frequently asked questions

Related guides

• How to save a Pandas DataFrame to a CSV file — the full df.to_csv() workflow: filenames, folders, columns, encoding.
• How to append to a CSV file with Pandas —mode="a" against mode="w", and the header trap.
• Common Pandas to CSV examples — ten worked calls, including this one, on the home page.