Although observers can discriminate visual targets with long exposures from otherwise-identical targets with shorter exposures, temporally overlapping distracters with an intermediate exposure can produce a striking degradation in performance. This new finding suggests that observers can only estimate one duration at a time. Discrimination on the basis of size, rather than duration, did not degrade as rapidly with the number of distracters but was still worse than predicted by unlimited-capacity models. The critical difference between estimates of temporal length and estimates of spatial length seems to be that the former can only be made at the end of an exposure, while the latter can be made at any time during an exposure. When sizes varied throughout the trial and decisions were based on terminal sizes, the set-size effect was as large as that obtained for duration discrimination. We conclude that when textural filters are not available for segregating a target from distracters, efficient estimates of size or duration require the serial examination of individual display items.