The repetition effect refers to the finding that reaction times are faster on trial n when the stimulus presented and/or the response required is the same as on trial n − 1 than when it is different. Five experiments examined the importance of stimulus features and response features in obtaining the repetition effect. Experiment 1 demonstrated a need for the stimuli to be categorically mapped to responses for a response repetition effect to be observed. Experiments 2–5 showed that the repetition effect can be obtained across responding hands when spatial information (Experiments 2 and 4) or finger information (Experiments 2 and 5) is consistent across hands but not when these sources of information are eliminated (Experiment 3). The results are in agreement with expectations developed from salient-features coding and with the inclusive links hypothesis proposed by H. Pashler and G. Baylis (1991a).