isnan, _isnan, _isnanf

Tests if a floating-point value is a NaN ("Not a Number").


int isnan(
   /* floating-point */ x
); /* C-only macro */

int _isnan(
   double x

int _isnanf(
   float x
); /* x64 only */

template <class T>
bool isnan(
   T x
) throw(); /* C++ only */


The floating-point value to test.

Return value

In C, the isnan macro and the _isnan and _isnanf functions return a non-zero value if the argument x is a NaN; otherwise they return 0.

In C++, the isnan template function returns true if the argument x is a NaN; otherwise it returns false.


Because a NaN value doesn't compare as equal to itself or to any other NaN value, to detect one, you must use one of these functions or macros. A NaN is generated when the result of a floating-point operation can't be represented in IEEE-754 floating-point format for the specified type. For information about how a NaN is represented for output, see printf.

When compiled as C++, the isnan macro isn't defined, and an isnan template function is defined instead. It behaves the same way as the macro, but returns a value of type bool instead of an integer.

The _isnan and _isnanf functions are Microsoft-specific. The _isnanf function is only available when compiled for x64.


Routine Required header (C) Required header (C++)
isnan, _isnanf <math.h> <math.h> or <cmath>
_isnan <float.h> <float.h> or <cfloat>

For more compatibility information, see Compatibility.

See also

Math and floating-point support
_fpclass, _fpclassf
isfinite, _finite, _finitef