Defined in header <math.h>
#define isnan(arg) /* implementation defined */
(since C99)

Determines if the given floating point number arg is a not-a-number (NaN) value. The macro returns an integral value.

FLT_EVAL_METHOD is ignored: even if the argument is evaluated with more range and precision than its type, it is first converted to its semantic type, and the classification is based on that (this matters if the evaluation type supports NaNs, while the semantic type does not).


arg - floating point value

Return value

Nonzero integral value if arg is a NaN, ​0​ otherwise.


There are many different NaN values with different sign bits and payloads, see nan.

NaN values never compare equal to themselves or to other NaN values. Copying a NaN may change its bit pattern.

Another way to test if a floating-point value is NaN is to compare it with itself: bool is_nan(double x) { return x != x; }


#include <stdio.h>
#include <math.h>
#include <float.h>
int main(void)
    printf("isnan(NAN)         = %d\n", isnan(NAN));
    printf("isnan(INFINITY)    = %d\n", isnan(INFINITY));
    printf("isnan(0.0)         = %d\n", isnan(0.0));
    printf("isnan(DBL_MIN/2.0) = %d\n", isnan(DBL_MIN/2.0));
    printf("isnan(0.0 / 0.0)   = %d\n", isnan(0.0/0.0));
    printf("isnan(Inf - Inf)   = %d\n", isnan(INFINITY - INFINITY));

Possible output:

isnan(NAN)         = 1
isnan(INFINITY)    = 0
isnan(0.0)         = 0
isnan(DBL_MIN/2.0) = 0
isnan(0.0 / 0.0)   = 1
isnan(Inf - Inf)   = 1


  • C11 standard (ISO/IEC 9899:2011):
    • The isnan macro (p: 236-237)
  • C99 standard (ISO/IEC 9899:1999):
    • The isnan macro (p: 217)

See also

returns a NaN (not-a-number)
classifies the given floating-point value
checks if the given number has finite value
checks if the given number is infinite
checks if the given number is normal
checks if two floating-point values are unordered

© cppreference.com
Licensed under the Creative Commons Attribution-ShareAlike Unported License v3.0.