The effects of intertrial interval (ITI) and feature–target interval (FTI) on learning of discrete-trial operant serial feature negative (target+/feature → target−) discriminations were examined in two experiments with rats. In Experiment 1, the FTI was 10 s and the ITIs were 0.5, 1, 2, 4, or 8 min, and in Experiment 2, the FTI was 20 s and the ITIs were 1, 2, 4, 8, or 16 min. Discrimination performance was acquired more rapidly with longer ITIs and with shorter FTIs. The best predictor of acquisition performance was the ratio of ITI and FTI. This predictive relation broke down at the smallest ratio value, which did not support discrimination learning when the FTI was 20 s. Transfer of the feature's inhibitory control to a separately trained target cue was minimal in all conditions, regardless of ITI. The results were discussed in the context of occasion setting.