The formalin test is increasingly used as a model of injury-produced pain but there is no generally accepted method of pain rating. To examine the properties of various pain rating methods we established dose-response relations for formalin injected in the plantar surface of one hind paw, and the analgesic effects of morphine and amphetamine using the most frequently reported behavioural measures of pain (favouring, lifting, licking and flinching/shaking of the injured paw) and combinations of these. Licking, elevation and favouring of the injected paw showed a biphasic response at all formalin doses. Flinching varied in form across the time course of formalin, and the biphasic nature of the behaviour was not as apparent. In untreated rats all these behaviours were infrequent. Flinching and favouring were increased after injection of local anaesthetic into the paw but remained negligible relative to the effect of formalin. Grooming other than that directed to the injected paw was elevated in a dose-dependent manner by formalin. Intercorrelations between the behaviours were different for the initial response and the second phase. Correlational analysis indicated that no single behavioural measure was a strong predictor of formalin, morphine and amphetamine dose. A simple sum of time spent licking plus elevating the paw, or the weighted pain score of Dubuisson and Dennis (1977), were superior to any single measure (r ranging from 0.75 to 0.86). Addition of flinching and favouring to the combined pain score using multiple regression did not increase variance explained. Depending on the measure used, a sedative dose of pentobarbital produced apparent analgesia, hyperalgesia or no effect. The interphase depression of pain, as well as the analgesic effects of morphine and amphetamine, were all associated with increased motor activation. Power analysis indicated that using a moderate dose of formalin and a combined pain score gave the greatest power to detect differences in pain. It was also found that pain scores increase with ambient temperature and that rat strains may differ in formalin pain sensitivity.