Pigeons received food only if they took longer than a specified time to begin and complete a fixed ratio. In Experiment 1, ratios with shorter durations had no stimulus consequence; in Experiment 2, these ratios ended with a stimulus change. In both studies, the mean time to complete the ratio exceeded requirements of less than 30 sec, approximately matched requirements of 30 sec, and fell progressively short of matching thereafter. Variability increased together with the means. The various effects resembled those of temporal differentiation experiments involving single responses. Although both number of ratios and time separating successive food presentations increased along with ratio duration, control experiments showed that differential reinforcement of duration, rather than either form or reinforcer intermittency, accounted for the performance. Experiment 2 also studied the effects of adding a stimulus that signalled when the required time had elapsed. The stimulus produced durations that matched even the most stringent requirements. This precision was not maintained when the stimulus was removed. Temporal differentiation schedules seem to have similar effects regardless of the response class and temporal property involved.In a temporal differentiation schedule, the presentation of a reinforcing stimulus depends on the duration of behavior. The duration can be of an individual event such as the latency, duration, or interresponse time of a bar press (e.g., Skinner, 1938), or it can refer to an entire sequence, such as the initial pause in a fixed-ratio schedule (Findley, 1962;Kelleher, Fry, and Cook, 1964) or the time taken to traverse an alley (Logan, 1960). Parametric studies involving both individual responses and sequences have demonstrated that performance adjusts to temporal requirements under a range of criterion values (Catania, 1970-response latency;Malott and Cumming, 1964; Staddon, 1965-interresponse time; Platt, Kuch, and Bitgood, 1973-bar-press duration; Zeiler, 1970Zeiler, , 1972-the duration of the fixedratio sequence). The data suggest that the effects of a temporal differentiation schedule may be independent of the particular behavior specified.