Inhibition of return (IOR) refers to slowed responding to targets at a location previously occupied by an irrelevant cue. Here we explore the interaction between stimulus-response (S-R) probability and IOR effects using go/no-go (Experiment 1) and two-choice discrimination tasks (Experiment 2). In both experiments, the IOR effect was larger for the likely S-R ensemble than for the unlikely one. In the first experiment, there were more false alarms for uncued targets than for cued targets, and this difference was larger for the unlikely S-R ensemble than for the likely one. In the second experiment, the same pattern was observed for incorrect keypress responses. As with voluntary orienting in response to predictive central cues, the results suggest that IOR affects late stages of processing by altering the criteria to respond to targets presented at the cued (previously attended) location.