Persistence of excitation extending beyond the offset of a stimulating sound has been proposed as a mechanism underlying forward masking. The persistence hypothesis was tested by comparing interaural level differences that led to a centered binaural image, between two partial-masking conditions, forward and the simultaneous masking. It was assumed that persistent excitation produced by a forward masker is equivalent to the excitation produced by an equally effective simultaneous masker. In the forward-masking condition, a 10-ms (5-ms rise/decay) 4-kHz signal was presented binaurally with a 5-ms delay after the offset of a monaural 150-ms 4-kHz masker. In the simultaneous-masking condition, the signal was added in the 90-deg phase during the final 10 ms of the steady-state portion of the same masker. The levels of the maskers were adjusted to produce the same masked threshold. The level of the signal in the masked ear was set to 10, 20, and 30 dB above the masked threshold. Listeners adjusted the level of the 10-ms signal in the unmasked ear until it produced a centered image. Under simultaneous masking, the centered image was obtained when the levels of the 10-ms tone were approximately equal in both ears. Under forward masking, the result depended on the level of the signal in the masked ear. For the signal presented 10 dB above the masked threshold, the level of the tone in the unmasked ear needed for a centered image was about 20 dB lower than that in the masked ear, suggesting strong attenuation of the response to the signal by the forward masker. The difference between the levels of the 10-ms tones in both ears decreased as the level of the partially masked signal increased. The observed differences between forward and simultaneous masking suggest that the two types of masking cannot be accounted for by the same mechanism and that forward masking is better described as attenuation rather than persistence. [Work supported by Grant No. DC00683 from NIDCD].