This study assessed the effectiveness of operant conditioning in training three species of captive callitrichid primates (Leontopithecus rosalia, Callithrix geoffroyi, and Saguinus imperator) to urinate on demand. There were three goals to the study: 1) to develop a system for quantitatively assessing positive reinforcement training; 2) to ascertain whether or not positive reinforcement techniques can be used to train callitrichid monkeys to urinate on demand, and if so, how many training sessions are required; and 3) to determine the effect on urination behavior of the trainer entering the cage to collect a urine sample. Positive reinforcement with a continuous reinforcement schedule was used to capture a natural behavior: urination. Training sessions (30 min each) were conducted at dawn thrice weekly during five consecutive phases: habituation, control, training (animals were rewarded for urinating), maintenance (animals had reached a defined training criteria and continued to be rewarded for urinating), and collection (animals were rewarded for urinating, and the trainer entered the cage to collect the sample). The numbers of 30-min training sessions required to train the three monkey species (L. rosalia, C. geoffroyi, and S. imperator) were five, six, and eight, respectively. For the three species, the mean number of urinations per animal was significantly greater during the training, maintenance, and collection phases compared to the control phase. However, the three species differed significantly in the manner in which the rates of urination changed across the five phases. A higher proportion of subjects urinated during the training, maintenance, and collection phases compared to the control phase. Latency to first urination varied significantly across the five phases, with significantly reduced latencies to urinate during the training, maintenance, and collection phases compared to the control phase. The entry of the trainer into the cage to collect the urine sample did not appear to alter urination behavior. We demonstrate that operant conditioning techniques, which typically incur minimal cost, time investment, and disturbance, can be used to increase the quantity of urine samples collected for physiological analysis, the proportion of animals that urinate, and the speed of sample collection. Am. J. Primatol. 64:83–93, 2004. © 2004 Wiley-Liss, Inc.