It has long been considered that lithium therapy could be abruptly stopped because it was guessed that lithium salts did not induce withdrawal symptoms. However, several open and controlled studies have shown that lithium discontinuation was associated with a possible withdrawal syndromee and with an incontestable rebound effect. Lithium withdrawal symptoms have been described in some patients, but it is not easy to distinguish them from depressive or manic symptoms, because no specific somatic withdrawal symptoms have been observed. The most important risk related to lithium discontinuation is the early recurrence of bipolar illness. Especially, it has been shown that the risk of manic recurrences is increased in the first weeks after discontinuation of lithium therapy, and that this risk is higher than the one predicted by the natural history of the manic-depressive illness. Relapses can occur even when lithium is stopped only for a few days. Abrupt discontinuation of lithium seems to be associated with an increased risk of recurrence. The pathophysiology of this rebound effect is still unknown. In clinical practice, lithium discontinuation has to be gradual when possible. Further studies are needed to precise at what time the risk of lithium withdrawal syndromee develops after starting lithium therapy. It is also necessary to establish more precise clinical guidelines for lithium discontinuation.