The goal of this study was to determine whether slowly infused, response-independent cocaine would reduce cocaine self-administration in an animal model of drug abuse. Seven male rhesus monkeys self-administered i.v. cocaine on a fixed-ratio 30 schedule (5-min time-out). With unit dose (0.056 mg/kg per infusion for one monkey and 0.032 mg/kg per infusion for the rest) and infusion volume (0.5 ml) held constant, the rate of delivery was manipulated (0.125, 0.1875, 0.375, 0.75 and 3 ml/min, with infusions lasting 240, 160, 80, 40, and 10 s, respectively). Response rates increased monotonically as a function of delivery rate. Responding for cocaine at the slowest delivery rate did not differ from saline. The effects of infusing additional cocaine (starting 30 min prior to the session) at this non-reinforcing rate (0.125 ml/min) were then determined. Delivery rate of the self-administered infusion was manipulated as before. Non-contingent cocaine significantly increased responding for cocaine (at the fastest delivery rate) and for saline. While non-contingent cocaine reduced responding for cocaine in two of the seven monkeys, it also significantly reduced responding in three monkeys that responded for food on the same schedule. Plasma levels of cocaine delivered at rates of 0.125 and 3 ml/min were compared in five other monkeys. While a higher peak was reached with the faster infusion, levels did not differ after 5 min. Thus, when an infusion became available (after the 5-min time-out) in the self-administration experiments, plasma levels should not have differed regardless of the delivery rate. These results suggest that a low-dose, slow-delivery treatment with cocaine might prime or reinstate drug seeking rather than decrease it.