Objective: To establish whether patients with a crisis plan had fewer voluntary or involuntary admissions, or fewer outpatient emergency visits, than patients without such a plan. Design: Multicenter randomized controlled trial with two intervention conditions and one control condition. Participants: Adult outpatients diagnosed with psychotic or bipolar disorder who had experienced at least one psychiatric crisis in the previous two years. Intervention: Two types of advance statement were used: (1) a crisis plan formulated by the patient with the help of a patient advocate (Patient Advocate Crisis Plan: PACP); and (2) a crisis plan developed together with the clinician (Clinician-facilitated Crisis Plan: CCP). Outcome: The percentages of patients admitted voluntarily or involuntarily (on an emergency basis or by court order), and the percentage who made outpatient emergency visits over an 18-month follow-up period. Results: A total of 212 patients were included: 69 in the PACP condition, 70 in the CCP condition, and 73 in the control condition. No effects of the two interventions were found on the numbers of voluntary admissions, involuntary admissions and emergency visits. Regarding involuntary admissions, there was no significant effect on emergency admissions, which were 17% (12/69) in the PACP condition, 10% (7/70) in the CCP condition, and 19% (14/73) in the control condition. There was a significant effect on planned court-ordered admissions, with 16% (11/69) in the PACP condition, 10% (7/70) in the CCP condition, and 26% (19/73) in the control condition. Finally, the interventions had no effect on outpatient emergency visits, with 32% (22/69) in the PACP group, 31% (22/70) in the CCP group, and 34% (25/73) in the control group. Conclusions: Crisis plans may be an effective intervention for reducing court-ordered admissions in patients with psychotic and bipolar disorders.