The program interruption that occurs when the result of a floating-point instruction has a nonzero fraction and a characteristic is smaller than the floating-point data format can handle. An exponent-underflow exception can be disabled by using the bit setting on a program mask.