Abstract : OBSTACLE-AVOIDANCE BY ECHOLOCATING LONG-EARED BATS (Plecotus townsendii) was tested while the bats were flying in random noise. Previous experiments had indicated that bats could achieve a surprising degree of success when dodging wires in a room filled with random noise in the band 20 to 50 kc, which covers the fundamental frequencies in the pulses of orientation sound used by Plecotus. The present experiments show that at a given noise power per cycle, the effectiveness of masking was far greater for a 10- to 90kc noise than for one that merely covered the fundamental frequencies in the bat pulses, but that the bats can still echolocate when the noise power, per cycle of bandwidth, exceeds the energy in the echo. Further broadening of the noise spectrum, to an upper limit exceeding 120 kc, had no discernible effect. The effect of coherent noise, originating in a single randomnoise generator connected to all the loudspeakers, was compared with the effect of noncoherent noise, which originated in four noise generators, each of which supplied one-quarter of the loudspeakers. In the noncoherent noise field, a bat always received noise from at least two independent noise generators. Though the bats seemed to fly with more assurance in coherent than in noncoherent noise, obstacle avoidance in the coherent noise was only slightly and inconsistently better.