Lithium, an important pharmacological agent for the treatment of manic-depressive illness in humans, is known to lengthen the circadian period in a number of different species. Recent experiments, on the other hand, suggest that pharmacological agents may affect the circadian system indirectly through an increase or decrease of activity. To explore the interaction between pharmacological and activity effects on the circadian system, lithium was administered chronically to three different strains of rats (ACI, BH, and LEW) while wheel-running activity was studied quantitatively. Two of these inbred strains (BH and LEW) show profound abnormalities in their circadian activity rhythms, namely, a reduced overall level of activity and bimodal or multimodal activity patterns. Wheel-running activity was monitored for 4 weeks under baseline conditions, followed by 3 weeks with lithium treatment (0.3% Li2CO3 administered with food) and 4 weeks with normal food. Treatment with lithium (average intake per day = 3.6 +/- 0.2 mg) consistently decreased both the overall level and the circadian amplitude of the activity rhythm. The free-running period tau was slightly lengthened during lithium treatment, while the most dramatic effect on period was observed after lithium withdrawal. Correlation analysis, however, revealed only a small negative correlation between activity level and period length, which proved significantly only for animals of the ACI strain. Our data support the traditional interpretation that lithium lengthens circadian period by a direct pharmacological effect on the circadian pacemaker rather than through indirect effects of activity feedback.