A time-sampling method that allows up to eight rats to be tested simultaneously in the formalin test is described and compared to the continuous rating method. Time sampling the behavioural response to formalin every 1 or 2 min produces scores that are essentially identical to continuous rating for both the formalin concentration effect relationship and the morphine dose effect relationship, with no loss of statistical power. The most important advantage of the method is that it allows data on other aspects of the rats' behaviour, such as behavioural state and the side effects of drugs to be scored during the formalin test. Formalin injection produces a dose-dependent decrease in locomotor and exploratory activity. The activity pattern of rats is normalized at morphine doses that produce about a 50% reduction in pain, while morphine doses high enough to completely suppress the pain response are accompanied by considerable sedation. The use of the jackknifing procedure to obtain unbiased estimates of the variability of parameters estimated from dose effect relationships is also described.