Clinicians and policymakers are examining with great interest the so-called "offset effect" of psychotherapeutic intervention. The study reported in this article attempts to examine more closely than previous studies the relationship between psychotherapeutic outcome and changes in health cafe utilization. Our data strongly imply that dropouts-patients who might be considered as having the "worst"psychotherapeutic outcomes-also have the "best" medical offset outcomes. These findings have major implications for those who would "sell" psychotherapy on the basis of its being cost-effective.Since the now classical FoIIette and Cummings study in 1967, which indicated that utilization of psychiatric services was associated with major decreases in subsequent medical care utilization for mental health patients, interest has grown in "medicaloffset" research. At this point, 14 years, two reviews (Jones & Vischi, 1979; Mumford, Schlesinger & Glass, 1981), and 15 studies later, the phenomena of medical offset after psychotherapy is still barely understood. Indeed, although the data only seem to indicate the offset potential of brief therapy, this same offset argument has been used to support the cost-effectiveness of long-term therapy and even psychoanalysis (Group for the Advancement of Psychiatry [GAP] 1978;Sharfstein, 1978).The authors of this article hypothesize that when psychotherapy research data, like the offset research referred to above, is indiscriminantly applied, we then risk sawing off the limb on which we sit. 200