Rats were runway trained with four different series of reward events for the purpose of determining if two different number cues (two and three) could be adequately employed to correctly anticipate a terminal nonreward in the series. In two of the series, two rewarded runs always occurred prior to terminal nonreward; one of these series always began with a nonrewarded run, NRRN, and the other did not, RRN. For two other series, three rewarded runs predicted terminal nonreward, NRRRN and RRRN. Each rat experienced each of the four series once a day for 26 days. The interval between runs within a series was about 20 sec, the interval between series was about 15 min, and the order in which the series were presented was randomized daily. Distinctive floor cues (smooth/black or rough/white, balanced over animals) signaled two-count and three-count series. Our animals correctly anticipated (ran slowly to) the terminal nonreward of all four series. On Day 27, an initial nonrewarded run was added to the two series that had been initiated by nonreward