Four food-deprived squirrel monkeys were trained to emit complex sequences of responses. The sequences involved pressing lighted response keys in orders dictated by colors that illuminated the keys, and ranged in length from two to five responses. Appropriate completion of these behavioral chains could be followed by food presentation. Acute administration of a range of doses (0.1-1.7 mg/kg) of cocaine hydrochloride produced dose-related decreases in the rate of completing chains and in accuracy of performance during chains. There was little evidence that the drug's effects on overall accuracy were related to the length of the chain. Three of the monkeys were exposed to daily administration of a large dose of cocaine, first after daily sessions and then prior to sessions. Daily postsession administration did not alter the dose-effect curves, but daily presession injection did, indicating the development of behavioral or "contingent" tolerance. In all cases, tolerance was accompanied by an increase in reinforcement frequency relative to the frequency observed following acute administration. Omission of the daily dose during presession drug administration resulted in performance near original control levels indicating essentially no withdrawal effect. The findings illustrate the importance of behavioral factors in the development of tolerance to cocaine in a primate.