The rotarod test of motor coordination in mice was modified by increasing the rotation speed every 30 sec until the animals fell off. This procedure yielded a stable, proficient level of performance within four brief trials; the approximately normal distribution of performance times provided an equivalent measure of either improvement or impairment caused by drugs and permitted the use of parametric statistical tests. A total of 240 mice were assigned to 20 different groups of 12 each, administered oral doses of placebo or different drugs, prior to trial 4. The use of a ratio score (performance time in trial 4 divided by the same animal's time in trial 3) provided a measure of drug-induced changes, controlling for individual differences among animals in over-all level of performance. Two phenothiazines (chlorpromazine and perphenazine) impaired performance at low doses, with a progressively greater decrement at increasing doses (4, 8, 16 mg/kg); 2 barbiturates (pentobarbital and amobarbital) showed an all-or-none effect, with no significant decrement at the lower doses (20 and 40 mg/kg) but almost complete incapacitation at the highest dose of 80 mg/kg. Performance superior to the placebo condition was found with the 2 lower doses of pentobarbital and with 3 d-amphet-amine doses (4, 8, 16 mg/kg). An analysis of individual differences gave evidence that the animals which were inferior in prior performance were more susceptible to both improvement and impairment of performance under the influence of drugs.