Evidence that outcome associations, position associations, and response patterns each contribute to performance in animal serial learning came from two experiments in which three-trial series of rewarded and not-rewarded trials were examined. Response patterns were disrupted in Experiment 1 by placing animals directly in the goal on selected trials, a procedure that reduced (favoring outcome associations) but did not eliminate (favoring position associations) serial behavior. Series constructed to make outcome associations impossible (Experiment 2) produced serial behavior that indicated salient position associations can serve as phrasing cues to chunk the position information contained in the series.