The ability to detect frequency changes in transposed sequences of tones was examined in a series of seven experiments. Listeners were asked to judge which of two transposed (i.e., frequency‐shifted) comparison patterns preserved the sequence of relative frequencies presented in a preceding standard pattern. The task was performed with five‐tone and two‐tone patterns under conditions of high and minimal pattern uncertainty. Regardless of pattern length or level of uncertainty, frequency discrimination thresholds for a change in the relative frequency of a single tone were considerably higher when patterns were transposed than when they were not. There was a tendency for performance to worsen with increasing degrees of transposition (primarily under high uncertainty) but most of the detrimental effects of transposition occurred within the first two semitones of transposition. Minimal uncertainty testing resulted in large improvements with five‐tone patterns (as much as one order of magnitude), but there was no effect of level of uncertainty on performance with two‐tone patterns. Thresholds for changes in two‐tone patterns were similar to (although slightly higher than) those for five‐tone patterns under minimal‐uncertainty testing. This pattern of results reveals that the effects of stimulus complexity (sequence length) and pattern familiarity (level of uncertainty) on relative‐frequency discrimination are quite similar to the effects of these variables on absolute‐frequency discrimination.