How to Check for NaN Values in Python

You will learn why NaN behaves differently from ordinary numbers, how to test for it with math.isnan(), and how to handle NaN safely in conditions, collections, and data-processing code.

NaN stands for Not a Number. It is a special floating-point value used to represent an undefined or invalid numeric result.

For example, a calculation may produce NaN when it cannot produce a meaningful number. Python can create it directly:

value = float('nan')

The reliable standard-library way to check a floating-point value is NaN is math.isnan():

import math

value = float('nan')
print(math.isnan(value))  # True

NaN matters because it does not follow normal comparison rules. In particular, NaN is not equal to itself. This unusual behavior can cause bugs in validation, filtering, and calculations unless code explicitly detects it.

Think of NaN as a numeric field containing an “unknown measurement” marker rather than an actual number.

A normal number can be compared with another number:

10 == 10  # True

But NaN means “there is no meaningful numeric value here,” so Python refuses to say that two NaN markers are equal—even if they appear to be the same:

nan = float('nan')
nan == nan  # False

Use math.isnan() as a dedicated scanner that asks: “Is this value the special unknown-number marker?”

Import the math module and pass a floating-point value to math.isnan().

import math

score = float('nan')

if math.isnan(score):
    print('The score is missing or invalid.')
else:
    print(f'The score is {score}.')

Output:

The score is missing or invalid.

math.isnan() returns a Boolean:

import math

print(math.isnan(3.14))           # False
print(math.isnan(float('nan')))   # True

A NaN value can also be detected with this property:

value = float('nan')
print(value != value)  # True

This works because NaN is the only ordinary Python floating-point value that is not equal to itself. However, prefer because its purpose is clear to readers.

Consider this validation function:

import math

def describe_temperature(temperature):
    if math.isnan(temperature):
        return 'Temperature is unavailable.'
    return f'Temperature: {temperature}°C'

reading = float('nan')
message = describe_temperature(reading)
print(message)

Step by step:

1. float('nan') creates a NaN floating-point value and stores it in reading.
2. describe_temperature(reading) calls the function with that value.
3. math.isnan(temperature) checks whether temperature is NaN.
4. The check returns True.
5. The function immediately returns 'Temperature is unavailable.'.
6. print(message) displays:

Temperature is unavailable.

With a valid value such as 22.5, returns , and the second runs instead.

NaN checks are useful whenever numeric input may be absent, invalid, or produced by a calculation.

• CSV and spreadsheet imports: A blank measurement may become NaN after loading data.
• Sensors and monitoring: A device may send an unavailable reading instead of a valid temperature, speed, or voltage.
• Financial calculations: Missing prices or rates should be identified before totals and averages are calculated.
• Web forms and APIs: Validate numeric data before storing it or using it in a calculation.
• Scientific computing: Filter invalid experimental results before graphing or computing statistics.

For example, skip invalid readings when calculating a total:

import math

readings = [12.5, float('nan'), 18.0]
total = 0

for reading in readings:
    if not math.isnan(reading):
        total += reading

print(total)  # 30.5

In production code, NaN checks are commonly placed near the boundary where data enters the program or just before important calculations.

Guard clause for validation

A guard clause handles invalid input early, keeping the main logic simple.

import math

def calculate_discount(price, percentage):
    if math.isnan(price) or math.isnan(percentage):
        raise ValueError('Price and percentage must be valid numbers.')

    return price * (percentage / 100)

Filtering a collection

import math

values = [10.0, float('nan'), 15.0, float('nan')]
valid_values = [value for value in values if not math.isnan(value)]

print(valid_values)  # [10.0, 15.0]

Replacing missing numeric values

Sometimes an application needs a fallback value rather than removing the record.

import math

def ():
     default  math.isnan(value)  value

(value_or_default(()))  

Comparing NaN with ==

This does not work:

value = float('nan')
print(value == float('nan'))  # False

Even a NaN value is not equal to itself. Use math.isnan(value) instead.

import math
print(math.isnan(value))  # True

Checking only against a variable containing NaN

This also fails:

missing = float('nan')
value = float('nan')

if value == missing:
    print('Missing')

Use math.isnan(value).

Calling math.isnan() on arbitrary text

math.isnan() expects a numeric value that can be treated as a float. Passing non-numeric text raises TypeError.

import math

value = float('nan')

math.isnan(value)       # True: preferred NaN check
math.isnan(4.2)         # False
value != value          # True for NaN, but less readable
value == float('nan')   # False: never use this to detect NaN

• NaN means “not a number.”
• NaN is a float value in Python.
• NaN is not equal to itself: nan == nan is False.
• Use math.isnan(value) for clear, reliable code.
• Check values before calculations if invalid readings must not affect the result.
• None is different from NaN; test it with value is None.

How do I check whether a value is NaN in Python?

Import math and use math.isnan(value).

import math
math.isnan(float('nan'))  # True

Why does float('nan') == float('nan') return False?

The IEEE floating-point rules define NaN as unequal to every value, including another NaN and itself.

Can I use value != value to detect NaN?

Yes. A NaN value is not equal to itself, so value != value is True. Prefer math.isnan(value) because it communicates the intent clearly.

Is NaN the same as None in Python?

No. NaN is a special floating-point value, while None represents the absence of a value. Check them with math.isnan(value) and value is None, respectively.

What happens if I add NaN to a number?

The result is usually NaN.

• Floating-point numbers — NaN is a special value defined by floating-point arithmetic.
• None and null values — Both can indicate missing data, but they require different checks.
• Input validation — NaN detection is a form of validating numeric input before use.
• Exceptions and ValueError — Invalid numeric data may be rejected with a clear error.
• Lists and list comprehensions — Useful for filtering NaN values from collections.
• Numeric comparison operators — NaN demonstrates why equality and ordering comparisons need careful handling.
• The math module — Provides common mathematical functions, including isnan().