Maintenance therapy in bipolar disorder (BD) is usually required to prevent relapses and improve residual symptoms. Therefore, in this study, we describe patterns of pharmacological maintenance treatment and identify associated clinical features. This prospective multicentre epidemiological study recruited a cohort of 739 consecutive out-patients with clinically stable BD. Clinical stability was assessed at baseline with the Clinical Global Impression scale for BD and depressive symptoms with the Hamilton Depression Rating Scale. Psychotropic medications were classified and analysed according to their mechanism as well as use. Logistic regression models were used to examine the associations between pharmacological strategies and clinical features. Longer time since last episode [odds ratio (OR) 1.002, p < 0.0001] and family history of psychiatric disorders (OR 1.911, p = 0.028) were associated with lithium in monotherapy; manic polarity of the most recent episode (OR 3.300, p = 0.006) and longer duration of clinical stability (OR 1.009, p = 0.034) with antipsychotic in monotherapy; depressive polarity of the most recent episode (OR 2.567, p = 0.003) and bipolar II disorder diagnosis (OR 2.278, p = 0.008) with antidepressant combination; no ongoing psychiatric co-morbidity (OR 0.230, p = 0.004) with lithium and anticonvulsant; manic polarity of the most recent episode (OR 3.774, p < 0.0001) with lithium and antipsychotic; manic polarity of the most recent episode (OR 2.907, p = 0.028) with lithium, anticonvulsant and antipsychotic. The pharmacological patterns followed published recommendations, except for the excessive use of antidepressants. This study reveals clinical factors closely related to prescription patterns.