Administration of 5% formalin into the rat or guinea pig hind paw evokes two spontaneous responses: flinching/shaking and licking/biting of the injected paw. The temporal and behavioral characteristics of these objective endpoints are described. Additionally, several practical suggestions aimed at standardizing this test for the evaluation of analgesics are presented. The early/acute and late/tonic (0-10 and 20-35 min post-formalin, respectively) phases of flinching were used to quantitate antinociception in the rat. PD 117302, the kappa selective agonist, was three times more potent than morphine against tonic flinching after SC administration. Formalin may therefore be a noxious stimulus of choice in the evaluation of kappa agonists. Morphine was only twice as potent against tonic flinching as against acute flinching or the tail-dip reflex to water (50 degrees C). In contrast, PD 117302 was 27 times less potent on early phase and was inactive in the tail-dip test. Thus, while morphine is essentially equipotent across tests, PD 117302 shows a spectrum of activity with impressive potency and efficacy being obtained against tonic pain. Kappa receptors may therefore be prominently involved in tonic pain states. Aspirin given orally was not consistently antinociceptive in either phase of the formalin test. Spinal transection completely abolished late phase responding but only partly attenuated flinching in the early phase. This suggests that the relative involvement of spinal (as opposed to supraspinal) processing of noxious inputs may, at least in part, be a function of stimulus intensity and underlie the differences in antinociceptive potency observed in this work.