In three experiments, musically trained listeners tried to detect local deviations from a typical pattern of expressive timing in the initial measures of Chopin's Etude in E Major. The goal of the experiments was to neutralize structurally based perceptual biases ("timing expectations") that had been revealed previously in the detectability of local deviations from metronomic timing in the same music. In Experiment 1, which presented deviations from full-scale expressive timing, the perceptual biases were reversed, indicating that listeners' timing expectations were smaller than the typical timing variation in performance. Experiments 2 and 3 therefore presented deviations from a down-scaled timing pattern (25% or 10% of its original magnitude). Although neither experiment succeeded in completely neutralizing the perceptual biases, correlational evidence suggested that listeners' timing expectations were about five times smaller than typical expressive timing variations. The small, essentially subliminal size of the listeners' timing expectations is consistent with recent findings of related unintentional timing variation in deliberately metronomic performance and in finger-tap synchronization with a metronomic rendition of the same music.