Kappa opioid agonists attenuate some neurochemical and behavioral effects of cocaine and are being considered as potential treatments for cocaine dependence. The present study examined the effects of two kappa opioid agonists, the benzomorphan ethylketocyclazocine (EKC) and the arylacetamide U50,488, on cocaine self-administration in rhesus monkeys. Monkeys responded for 0.032 mg/kg/injection cocaine (i.v.) and 1 g banana-flavored food pellets during alternating daily sessions of cocaine and food availability. Chronic treatment for 10 consecutive days with EKC (0.0032-0.032 mg/kg/hr) or U50,488 (0.032-0.1 mg/kg/hr) dose-dependently decreased self-administration of cocaine unit doses at the peak of the cocaine dose-effect curve (0.01 and 0.032 mg/kg/injection). These decreases in cocaine self-administration were often sustained throughout the 10 days of treatment. Doses of EKC and U50,488 that decreased cocaine self-administration usually decreased food-maintained responding as well. In addition, EKC and U50,488 often produced emesis and sedation during the first few days of treatment, although tolerance appeared to develop rapidly to these effects. In general, EKC produced fewer undesirable effects than U50,488 at doses that decreased cocaine self-administration. The kappa antagonist norbinaltorphimine (3.2 mg/kg) did not affect responding maintained by cocaine or food. However, both norbinaltorphimine (3.2 mg/kg) and the opioid antagonist naloxone (1.0 mg/kg/hr) blocked the effects of EKC and U50,488. These results indicate that chronic administration of EKC and U50,588 produce a dose-dependent, kappa receptor-mediated and often sustained decrease in cocaine self-administration. However, these kappa agonists also produce undesirable behavioral effects that may complicate their use as treatments for cocaine dependence.