Polars · Module

df.drop_nulls

Removes rows containing nulls from a Polars DataFrame. By default every column is checked, but subset can restrict the check to named columns or a selector.

How it works

drop_nulls returns a new DataFrame after removing rows that have nulls in the selected columns. With no subset, every column participates; with a column name collection or selector, only those columns determine which rows are removed, and the surviving rows keep their original order.

Syntax

DataFrame.drop_nulls(subset: ColumnNameOrSelector | Collection[ColumnNameOrSelector] | None = None) → DataFrame

Parameters

Name	Type	Description
`subset`	`ColumnNameOrSelector \| Collection[ColumnNameOrSelector] \| None = None`	Column name or names whose null values are considered; when `None`, all columns are used. A selector is also accepted.

Returns

A new DataFrame with rows containing considered null values removed.

Key points

The method removes any row containing at least one null among the columns it considers.
Rows that survive filtering remain in their existing relative order.
Leaving subset unspecified makes every DataFrame column participate in the null check.
Providing subset limits the columns whose nulls can cause row removal.
The subset can be expressed with column names or a Polars selector.

Common mistakes

This method does not remove NaN values; use drop_nans() when NaN handling is required.

Example: Drop rows using a column selector

import polars as pl
import polars.selectors as cs

df = pl.DataFrame({
    "id": [1, 2, 3],
    "score": [10, None, 30],
    "label": ["a", "b", None],
})

clean = df.drop_nulls(subset=cs.numeric())
print(clean)

Removes rows with nulls in numeric columns while retaining the row whose only null is in the non-numeric label column.

When to use it

Remove incomplete rows

Prepare complete-case data before analysis when any missing field makes a row unusable.

Check selected fields

Validate required fields while allowing optional columns to remain null.

Filter with selectors

Apply the same missingness rule across a type-defined group of columns.

Interview questions

QWhat does DataFrame.drop_nulls() do by default?

It returns a DataFrame containing only rows without nulls in the columns being considered. If subset is omitted, all columns are checked, and the remaining rows preserve their original order.

Follow-up: How would you retain rows with nulls in nonessential columns? Pass only the required column names or a selector through subset, so nulls elsewhere do not trigger removal.

QHow does the subset parameter change behavior?

subset identifies the columns whose null values should cause a row to be dropped. It accepts one column name, a collection of column names, or a selector; None means all columns.

Follow-up: How could you apply the same rule to every integer column without listing them? Use a selector such as polars.selectors.integer() as the subset.

QDoes drop_nulls remove NaN values?

The operation checks for null values, not NaN values. A floating-point NaN therefore requires the separate drop_nans() method if it should be removed.

Follow-up: What should be used when both missing-value representations may occur? Apply the appropriate null and NaN operations explicitly, because the documentation distinguishes the two values.

DataFrame.drop_nans
polars.selectors.integer

Source: polars.DataFrame.drop_nulls — Polars documentation. Checked 2026-10-15.