Background: A substantial number of people with bipolar disorder show a suboptimal response to treatment. Aims: To study the effectiveness of a collaborative care programme on symptoms and medication adherence in patients with bipolar disorder, compared with care as usual. Method: A two-armed, cluster randomised clinical trial was carried out in 16 out-patient mental health clinics in The Netherlands, in which 138 patients were randomised. Patient outcomes included duration and severity of symptoms and medication adherence, and were measured at baseline, 6 months and 12 months. Collaborative care comprised contracting, psychoeducation, problem-solving treatment, systematic relapse prevention and monitoring of outcomes. Mental health nurses functioned as care managers in this programme. The trial was registered with The Netherlands Trial Registry (NTR2600). Results: Collaborative care had a significant and clinically relevant effect on number of months with depressive symptoms, both at 6 months (z = 72.6, P = 0.01, d = 0.5) and at 12 months (z = 73.1, P = 0.002, d = 0.7), as well as on severity of depressive symptoms at 12 months (z = 72.9, P = 0.004, d = 0.4). There was no effect on symptoms of mania or on treatment adherence. Conclusions: When compared with treatment as usual, collaborative care substantially reduced the time participants with bipolar disorder experienced depressive symptoms. Also, depressive symptom severity decreased significantly. As persistent depressive symptoms are difficult to treat and contribute to both disability and impaired quality of life in bipolar disorder, collaborative care may be an important form of treatment for people with this disorder.