The present study examined auditory backward masking in a task requiring subjects to label a target sound as having a long or short duration. Perception of target duration was influenced by the interaction of the effects of (1) the similarity of the target/mask duration, (2) the acoustic similarity of the target and mask (tone vs. white noise), and (3) the interstimulus interval (ISI) between the target and mask. Specifically, relative to the long duration (100 ms) mask, the short duration (40 ms) mask helped performance for the target having a similar short duration (55 ms) but hurt performance for the target having a dissimilar long duration (85 ms). This effect of the similarity of target/mask duration was greater for acoustically similar targets and masks than for acoustically dissimilar targets and masks, and particularly so at the intermediate (45, 105, 165, 205 ms) ISIs. These results can be explained within the framework of Massaro and Idson's (1976) two-stage model by assuming that masking of perceived auditory duration is the result of two processes: (a) the mask's interruption of target processing in a peripheral auditory processing stage, and (b) a confusion of which item is the target vs. the mask, due to temporal uncertainty in the transfer of the target and mask into a central auditory processing stage.