Pandas · Function

pd.to_datetime

Converts scalar, array-like, Series, or DataFrame/dict-like input into pandas datetime objects. Options control error handling, parsing formats, timezone normalization, numeric units, origin, and caching.

How it works

Use pd.to_datetime to convert scalar values, sequences, Series, or table-like data into pandas datetime objects. Parsing behavior can be controlled with format, errors, timezone normalization through utc, units for numeric values, and origin for numeric timestamps.

Syntax

pandas.to_datetime(arg, errors='raise', dayfirst=False, yearfirst=False, utc=False, format=None, exact=<no_default>, unit=None, origin='unix', cache=True)

Parameters

Name	Type	Description
`arg`	`int, float, str, datetime, list, tuple, 1-d array, Series, DataFrame/dict-like`	Value or collection to convert; supported inputs include numeric and textual values, datetime objects, one-dimensional arrays, Series, and DataFrame/dict-like data.
`errors`	`{'raise', 'coerce', 'ignore'}, default 'raise'`	Controls failures: raise an exception, coerce invalid values to NaT, or ignore conversion.
`dayfirst`	`bool, default False`	Interpret ambiguous dates with the day before the month.
`yearfirst`	`bool, default False`	Interpret ambiguous dates with the year first.
`utc`	`bool, default False`	Convert results to timezone-aware UTC values.
`format`	`str, default None`	strftime pattern used to parse strings.
`exact`	`bool, default <no_default>`	Require the complete format to match when a format is supplied.
`unit`	`str, default None`	Unit for numeric timestamps, such as D, s, ms, us, or ns.
`origin`	`scalar, default 'unix'`	Reference point for numeric timestamps.
`cache`	`bool, default True`	Use a cache for unique converted date strings.

Returns

The return type depends on the input: scalar input returns Timestamp, array-like input returns DatetimeIndex, Series input returns Series, and DataFrame input returns Series; Python datetime or object-dtype fallbacks can occur for unsuccessful timezone or out-of-range parsing.

Key points

It provides one conversion interface for individual values, collections, Series, and table-like data.
The required arg parameter may be an integer, float, string, datetime, list, tuple, one-dimensional array, Series, or DataFrame/dict-like object.
format accepts a strftime pattern so parsing can match a known string layout.
utc=True produces timezone-aware UTC output, including localization of timezone-naive values.
Caching is enabled by default and can improve conversion when inputs contain repeated date strings.

Common mistakes

With the default errors='raise', an invalid date string raises a ParserError; choose errors='coerce' when invalid values should become missing timestamps.
dayfirst influences ambiguous numeric dates but is not a strict requirement, so explicitly provide format when the input layout must be enforced.
Setting utc=True normalizes timezone-aware inputs to UTC and localizes timezone-naive inputs to UTC.

Example: Parse dates as UTC

import pandas as pd

values = pd.to_datetime(
    ["2024-01-15", "2024-02-20"],
    format="%Y-%m-%d",
    utc=True,
)
print(values)

Converts ISO date strings into a UTC-aware DatetimeIndex.

When to use it

Normalize date columns

A column contains date strings and needs datetime operations such as sorting, filtering, or extracting components.

Decode numeric timestamps

A dataset stores Unix timestamps as integers or floats.

Coerce invalid dates

Inputs may contain malformed date strings that should not stop a pipeline.

Normalize timestamps to UTC

Data from multiple sources uses mixed timezone conventions.

Interview questions

QWhat inputs does pandas.to_datetime support, and what does it return?

The function accepts scalars, array-like values, Series, and DataFrame or dict-like inputs. The result type follows the input: a scalar produces a Timestamp, an array-like produces a DatetimeIndex, and a Series produces a Series.

Follow-up: What happens for a DataFrame? It must provide at least year, month, and day columns, and the result is a Series of datetime values.

QHow can parsing failures be handled?

errors='raise' propagates parsing failures, errors='coerce' turns invalid values into NaT, and errors='ignore' leaves the input unchanged when parsing fails. The default is raise.

Follow-up: How would you make parsing deterministic for ambiguous strings? Supply an explicit strftime format rather than relying on inference.

QHow do timezone and numeric timestamp options affect conversion?

utc=True makes the output timezone-aware and UTC-normalized. Numeric inputs can be interpreted using a unit such as seconds or milliseconds and an origin, whose default is Unix time.

Follow-up: Why specify unit for integers? Without it, numeric timestamps are not given the intended time scale; unit identifies the timestamp resolution.

pandas.Timestamp
pandas.DatetimeIndex
pandas.to_timedelta
pandas.date_range
Series.dt

Source: pandas.to_datetime. Checked 2026-10-15.