Six rats were trained with food deliveries contingent upon their pressing a lever and holding it down for fixed, cumulative durations. Hold requirements were varied from 7.5 seconds to 120 seconds. Lever holding was maintained reliably at hold requirements as long as 30 seconds to 105 seconds for different rats. At longer hold requirements, lever holding was erratic and tended to occur only early in sessions. At shorter and intermediate requirements, the patterns of lever holding resembled those of responding under fixedratio schedules for discrete responses, with breaks in responding immediately after reinforcement alternating with relatively continuous lever holding until the next reinforcement. At longer hold requirements, postpause lever holding frequently was interrupted with additional pauses. The duration of postreinforcement pauses increased linearly with the scheduled hold requirement. However, for five of six rats, the hold requirement, which represents the actual time spent lever holding per reinforcer, accounted for somewhat less variance in pause duration than did interreinforcement time.Key words: postreinforcement pause, interreinforcement time, work time, continuous response, lever holding, rats Performance under schedules that provide reinforcement periodically, such as fixedinterval (FI) and fixed-ratio (FR) schedules, is characterized by a period of time immediately following reinforcement during which no responding occurs. The duration of this postreinforcement pause increases monotonically as the fixed interval (e.g., Harzem, 1969;Innis & Staddon, 1971;Lowe & Harzem, 1977;Schneider, 1969;Shull, 1970Shull, , 1971Skinner, 1938;Wilson, 1954) pause durations were approximately the same under FR schedules and yoked-interval schedules in which interreinforcement times matched those obtained from the FR schedules. Nevin (1973) analyzed data obtained by Berryman and Nevin (1962) and found that pause duration was a linear function of the average interreinforcement times obtained under FI, FR, and interlocking FI FR schedules. Rider (1980) found that pause duration was linearly related to average interreinforcement times obtained under alternative FI FR schedules over a broad range of schedule parameters.The good linear fit between pause duration and interreinforcement time across simple Fl and FR schedules and complex interlocking and alternative schedules raises the possibility that interreinforcement time controls pausing independently of the particular schedule of reinforcement. However, in a direct comparison of FI and FR schedules with comparable interreinforcement times, Capehart, Eckerman, Guilkey, and Shull (1980)