First, the control patients obtained medication from different sources without their psychiatrists’ knowledge. We are attempting to trace the sources of these prescriptions, but this is a challenge even within the context of a socialized mental health care system. Second, we have no way of knowing the level of patient compliance with the medication regimen, particularly in the control group. During the follow-up period three patients in the control group had prescriptions for lithium, but blood drug levels, when they were available, suggested that compliance was poor. Of course, we had no serum levels for the patients taking other forms of medication. Nonetheless, we agree with the points suggested by Dr. Stern, and we will try to report more specific details of the prescribing patterns in a future report on cost-effectiveness. Finally, we are engaged in an ongoing, randomized, controlled outpatient treatment study of patients with borderline personality disorder without previous partial hospitalization.