Previous research on echo detection in bats has suggested that the effective threshold is a function of the acoustic clutter in the experimental environment, as might be expected given the low ambient noise levels typical of such psychophysical research. This paper demonstrates that theory of signal detectability (TSD) methodology is applicable to bats and uses it to show that an important element of clutter limiting in Eptesicus fuscus and Noctilio leporinus is backward masking of phantom targets by the real echo from the loudspeakers used to generate them. This information suggests that a previous estimate of the magnitude of automatic gain control (AGC) is too high, due to variable backward masking inherent in the experimental method used. A re-examination of gain control using a masking-free method shows that it reduces auditory sensitivity by 6 to 7 dB per halving of target range, rather than 11 dB as previously thought.