Although research on pathological gambling is increasing, there are few studies examining correlates of treatment retention. In the current study, 50 outpatients with a primary DSM-IV diagnosis of pathological gambling treated in a clinical practice were assessed by chart review. Standard scales were used to rate subjects at baseline and at 2-month intervals. Subjects who dropped out of treatment were contacted by telephone to determine reasons for discontinuation of treatment. The mean duration of follow-up was 360.4 +/- 352.2 days. Twenty-four (48%) of the subjects discontinued treatment, including 14 (36.8%) of those subjects who were responders (defined as Clinical Global Impression [CGI] scale of "very much" or "much" improved) during at least one assessment point. Of those who discontinued treatment, 41.7% reported missing the thrill of gambling and 20.8% reported feeling certain that they could win and relieve financial burdens. Predictors of treatment continuation were responding to treatment within 8 weeks (odds ratio [OR], 6.00; 95% confidence interval [CI], 1.13 to 32.00; P =.04) and having a supportive environment (OR, 22.99; 95% CI, 5.04 to 104.76; P <.001). We conclude that a large percentage of patients with pathological gambling discontinue treatment. Predictors of treatment continuation may have clinical importance. Prospective longitudinal studies are needed to further elucidate the course of pathological gambling.