It is currently agreed that starting a lithium therapy is justified, as soon as the evolution of recurrent affective disorder is evident, by: the occurrence of an acute depressive episode during the four years preceding the current acute episode, that is to say 2 obvious acute episodes within 5 years which is enough to make highly probable a new acute short or medium dated episode; an acute manic episode caused by the high level of recurrence in bipolar patients and the frequent severe congruencies of an acute manic episode. The prophylactic efficacy of the lithium salts has been proved by several placebo-controlled studies: the result of these trials states that with placebo the recurrence level varies between 38% and 93%, whereas with lithium it is between 0% and 44%. The controlled trials in the sixties and seventies have made evident that about 1/3 of bipolar patients do not respond to lithium, the results varying according to the selection tests, the duration of the observation, the definition of the failure. We'll address three up to date questions. 1) Whereas the level of the lithium therapy is rather well established (0.6 to 1 mEq/l) and is of a great preventive efficacy with patients that are maintained in the high zone of the therapy scale, the present question is to know if it is necessary to split up the dose of lithium or to administer the daily dose in a single dose in the evening.(ABSTRACT TRUNCATED AT 250 WORDS)