pandas KeyError: not in index / None of [Index] (tested)

KeyError: "['x'] not in index" means some of the columns (or row labels) you selected do not exist; "None of [Index([...])] are in the [columns]" means none do; "not found in axis" comes from drop(). We reproduced all three on pandas 2.3.3 and 3.1.0.

1. df[[...]] with a missing column

import pandas as pd

df = pd.DataFrame({"a": [1, 2], "b": [3, 4]})
for cols in (["a", "x"], ["x", "y"]):
    try:
        df[cols]
    except KeyError as e:
        print("KeyError:", e)

print(df[df.columns.intersection(["a", "x"])].columns.tolist())   # keep only existing
print(df.reindex(columns=["a", "x"]).values.tolist())             # missing -> NaN
missing = set(["a", "x"]) - set(df.columns)
print("missing:", missing)

Output with pandas 2.3.3:

KeyError: "['x'] not in index"
KeyError: "None of [Index(['x', 'y'], dtype='object')] are in the [columns]"
['a']
[[1.0, nan], [2.0, nan]]
missing: {'x'}

Output with pandas 3.1.0:

KeyError: "['x'] not in index"
KeyError: "None of [Index(['x', 'y'], dtype='str')] are in the [columns]"
['a']
[[1.0, nan], [2.0, nan]]
missing: {'x'}

Selecting a list of columns raises "['x'] not in index" when some are missing (it lists only the missing ones) and "None of [Index([...])] are in the [columns]" when none exist. The only version difference is the dtype in the message: pandas 3 stores strings as str, pandas 2 as object. Find the missing names with a set difference, then either fix the name (see column exists but KeyError for spaces and BOMs), keep only existing columns with df.columns.intersection(...), or use reindex(columns=...) to get NaN columns instead of an error.

2. drop(): "not found in axis"

import pandas as pd

df = pd.DataFrame({"a": [1, 2], "b": [3, 4]})
try:
    df.drop("x", axis=1)
except KeyError as e:
    print("KeyError:", e)
try:
    df.drop("a")                     # default axis=0 looks in the row index
except KeyError as e:
    print("KeyError:", e)

print(df.drop(columns="a").columns.tolist())
print(df.drop(columns=["a", "x"], errors="ignore").columns.tolist())

Output (identical on pandas 2.3.3 and 3.1.0):

KeyError: "['x'] not found in axis"
KeyError: "['a'] not found in axis"
['b']
['b']

df.drop("a") fails even though a is a column, because drop defaults to axis=0 (row labels). Write df.drop(columns="a"). To drop columns that may not exist, add errors="ignore".

3. Row labels in .loc, and set_index

import pandas as pd

df = pd.DataFrame({"a": [1, 2]}, index=[10, 20])
try:
    df.loc[[10, 30]]
except KeyError as e:
    print("KeyError:", e)
try:
    df.set_index("x")
except KeyError as e:
    print("KeyError:", e)

Output (identical on pandas 2.3.3 and 3.1.0):

KeyError: '[30] not in index'
KeyError: "None of ['x'] are in the columns"

The same "not in index" message appears for row labels in df.loc[[...]]; check with df.index.isin([...]) or select with df[df.index.isin(labels)]. set_index on a missing column says "None of ['x'] are in the columns" (without the Index wrapper).

Related: KeyError: 0 in pandas · Error tokenizing data · SettingWithCopyWarning.

Written with AI assistance. Measured on 2026-10-23 in the python:3-alpine Docker image (Python 3.14) with pandas 2.3.3 and pandas 3.1.0. Each block was run verbatim on both versions and the output pasted unedited; where the versions differ both outputs are shown. PlainHeaders is our extension; Python and pandas are free.