Dichotic sequence discrimination was studied by means of a three-alternative forced-choice method. Two of the alternatives ('same') consisted of a single pair of dichotic clicks with the stimulus at one ear preceding the stimulus at the other ear by a given temporal separation (e.g., R-Δt-L). One of the alternatives consisted of a single pair of clicks with the opposite sequence (e.g., L-Δt-R). The subjects' task was to discriminate between the 'different' and the 'same' dichotic pairs. The results indicate that discrimination is a nonmonotonic function of Δt, i.e., V- or U-shaped, with high levels at both short and long temporal separations (Δt=2 and 4 ms and Δt=64 and 96 ms) and with lower levels of discrimination for intermediate Δts. A second experiment was performed to study the effect of continued practice by subjects trained in dichotic sequence discrimination for Δts 4 and 64 ms (high discrimination levels) and for Δt=16 ms (low discrimination level). There was no systematic trend towards improvement over 8 blocks of 50 trials for each Δt. Discrimination levels of these Δts did not differ between Experiments 1 and 2.