Two rats were trained to detect brief 8000-Hz tones centered in a one-third octave band of noise. The procedure was analogous to the yes-no method of human psychophysics in that one response was defined as correct and reinforced if the tone were present in the noise, and another response was correct and reinforced if the tone were absent. The percentage of correct responses was determined principally by the energy in the tone for the range of durations studied (75 to 600 msec): if the tone's duration were halved, for example, its power had to be doubled to keep the percentage of correct responses about the same. The ratio of the energy in the tone to the power per cycle of the noise needed to maintain 75% correct responses was about 36 db for one animal and 41 db for the other. Although the two responses were similar, and their consequences equal, biases in responding were sometimes observed.