Until recently, the rate at which patients switch from bipolar depression to the manic or hypomanic phase of the disorder during treatment with antidepressant medications was poorly defined. The completion of three large-scale, double-blind controlled trials in bipolar I depression has improved understanding of this phenomenon. The low switching rates observed in these studies of lamotrigine, paroxetine and moclobemide may indicate a special application of these drugs in the management of patients prone to antidepressant-induced switching. These studies also confirm prior suggestions that tricyclic antidepressants present the highest risk of switching. At present there is no consensus over the optimal definition of switching. Standardising the definition may lead to improvements in the clinical management of bipolar disorder.