Until recently it appeared that lithium carbonate possessed a unique spectrum of clinical action in the acute and prophylactic treatment of manic and depressive episodes. It is now increasingly apparent that the anticonvulsants carbamazepine and valproate also share components of this spectrum of efficacy in the affective disorders but, in addition, are clinically effective in some lithium nonresponders. This clinical convergence can now drive a reexamination of the potential mechanisms of action of these compounds in the affective disorders. In spite of intensive study over several decades, the mechanism of action of lithium has remained elusive. A basic conundrum in the consideration of the actions of lithium has also been to explain how a simple ion could have such complex effects on multiple neurotransmitter systems and, in particular, have bimodal actions in the treatment of both manic and depressive phases of the illness. We suggest that a fundamental reconceptualization of both mania and depression as overactivated neural systems (either excitatory or inhibitory) could facilitate this conceptualization. Given the recent evidence linking lithium's effects to uncoupling receptor-mediated activity at the level of G-proteins or attenuating it at the level of second messenger systems mediated by adenylate cyclase or phosphoinositide turnover, these mechanisms become ideal candidates for considering how the drug could dampen overactivated systems potentially relevant to either depression or mania. The lag in onset of maximum therapeutic action of lithium, carbamazepine, and valproate further suggests that biologic effects associated with chronic compared with acute administration are the prime candidates for psychotropic effects. Comparison of the acute and chronic effects of carbamazepine with those of valproate is also offered to focus on the most likely receptor, second-messenger, and ion channel mechanisms involved in their anticonvulsant and psychotropic actions. It is hoped that better understanding of the comparative actions of lithium, carbamazepine, and valproate will allow better targeting of individual drugs for individual patients as well as, ultimately, the development of new and more selective treatments for the recurrent affective disorders.