In the 1970s, several randomized controlled trials demonstrated significant antimanic and antidepressant properties of lithium in the prophylactic treatment of bipolar disorder. However, a recent meta-analysis of randomized, placebo-controlled trials of lithium in bipolar disorder found that its protective effect against depressive relapse/recurrence was equivocal. By examining potentially relevant parameters of recent randomized controlled trials with regard to lithium's prophylactic antidepressant efficacy, we try to identify factors which might help to explain these discrepant results across the different trials. Lithium's efficacy against manic relapse/recurrence appears rather robust at plasma levels between 0.8 and 1.2 mmol/L, whereas lithium's efficacy against depressive relapse/recurrence may be more modest and dependent on whether a response during the preceding acute episode was achieved by lithium treatment. Furthermore, it might be advisable to continue lithium without interruption at the same dose/plasma level, which yielded the initial response. A lithium level between 0.5 and 0.8 mmol/L may be equally efficacious against overall relapse and associated with equal or even superior efficacy regarding depressive relapse/recurrence. To provide evidence-based guidelines on this issue, large prospective, randomized, double-blind, placebo-controlled trials are needed comparing the efficacy of lithium at different plasma levels against manic and depressive relapse/recurrence. In these trials, factors previously associated with predicting response to lithium should also be assessed.