Fixed‐interval performances of rats were described either in terms of the individual intervals of the session or in terms of a single average interval constructed for the entire session. Responding in the individual intervals usually followed break‐and‐run and single response patterns rather than the scalloped pattern that emerged when the results were averaged. There was, however, a reasonable correspondence between the quarter‐life values calculated from individual intervals and those calculated from the averages. According to the pattern exhibited by the average interval, the probability of a response increased as the interval elapsed. The same conclusion was indicated by more molecular analyses of the conditional probabilities of pause terminations. The results showed that descriptions of fixed‐interval data in terms of overall averages reveal aspects of performance that are not immediately apparent within individual