Using a target simulator three serotine bats, Eptesicus serotinus, were trained to judge whether a phantom target was present or absent. The echolocation sounds emitted by the bats during the detection were intercepted by a microphone, amplified and returned by a loudspeaker as an artificial echo, with a delay of 3.2 ms and a sound level determined by the overall gain and cry amplitude. The cry level of each pulse was measured and the echo level received by the bat was calculated. The target was presented in 50% of the trials and the gain adjusted using conventional up/down procedures. Under these conditions between 40 and 48 dB peSPL were required for 50% detection (Figs. 2, 3). In a subsequent experiment the phantom target was masked with white noise (No) with a spectrum level of -113 dB re. 1 Pa X Hz-1/2. The thresholds were increased by 7-14 dB. Energy density (S) of a single pulse was measured and used to estimate S/No, which ranged from 36-49 dB at threshold. Theoretically the coherent receiver model predicts the ratio between hits and false alarms observed for the bats at a S/No of ca. 1-2 dB. Since the bats require 40-50 dB higher S/No (Fig. 3), this is taken as negative evidence for coherent reception (cross correlation). Furthermore, a strong sensitivity to clutter was found since there seemed to exist a fixed relationship between thresholds and clutter level.