Treatment of bipolar disorder (BD) has traditionally focused on alleviation of acute symptoms and prevention of future recurrences. Current treatment guide-lines advocate more or less similar treatment algorithms for all patients. Such approach largely ignores the clinical, genetic, and pathophysiological heterogeneity of BD, which makes certain patients more (or less) likely to respond to specific treatments. Variables such as family history, comorbidity, course of illness, quality and duration of previous remissions, physical and medical comorbidity, and side-effects may help in selecting the most effective treatment for an individual patient, yet their value is not recognized by current algorithms. As well, polymorphisms of specific genes may prove useful in predicting treatment outcome and/or understanding the pharmacological mechanisms of mood stabilization. Novel molecular targets have recently emerged from studies of mechanisms of action of available mood stabilizers. They include inhibitors of protein kinase C, inhibitors of glycogen synthase kinase, or medications modulating glutamatergic neurotransmission. As well, treatment targets are moving beyond acute symptoms and prevention of mood episodes. Cognitive deficits, persistence of residual symptoms, and increased mortality of BD are recognized as important for outcome of BD, yet are not always adequately addressed by traditional treatments.