Kappa opioid agonists inhibit dopamine release from mesolimbic dopaminergic neurons and attenuate some behavioral effects of cocaine in rodents. Evidence that kappa opioid agonists may act as functional antagonists of cocaine led us to examine their interactions with cocaine's abuse-related effects in rhesus monkeys. In cocaine self-administration studies, four arylacetamides (U50,488, enadoline, (-) spiradoline and PD117302) and four benzomorphans (ethylketocyclazocine [EKC], bremazocine, Mr2033 and cyclazozine) each were administered as continuous infusions over 10 days. EKC, Mr2033, bremazocine, U50,488 and enadoline produced significant dose-dependent and sustained decreases in cocaine self-administration and also decreased food-maintained responding at some doses. Emesis and sedation were occasionally observed during the first two days of kappa agonist treatment, but tolerance developed rapidly to these effects. Cyclazocine, PD117302 and spiradoline did not significantly alter cocaine self-administration. The behavioral effects of EKC and U50,488 were antagonized by both the kappa opioid antagonist nor-binaltorphimine and the non-selective opioid antagonist naloxone. In general, compounds with mixed activity at both kappa and mu opioid receptors (e.g. EKC, Mr2033) decreased cocaine self-administration more consistently and with fewer or less severe undesirable side effects than more selective kappa agonists (e.g. U50,488, spiradoline). Although several kappa agonists decreased cocaine self-administration, EKC and U50,488 did not consistently block the discriminative stimulus effects of cocaine in monkeys trained to discriminate cocaine from saline. The extent to which kappa agonist-induced decreases in cocaine self-administration reflect an antagonism of cocaine's abuse-related effect remains to be determined.