Dysthymia is one of the most prevalent problems in primary care, especially in the elderly. In this study, we evaluated the demographic and clinical predictors of nonresponse to treatment in primary care patients with dysthymia. The study sample consisted of 338 primary care patients meeting DSMIII-R criteria for dysthymia from 4 diverse geographic sites in a randomized controlled 11-week trial of paroxetine, problem-solving therapy or placebo. Patients who attended at least 4 treatment sessions were used in the analysis. A score of less than 7 on the Hamilton was defined as a positive response to treatment. By Week 11, 52.2% of patients had a positive response to treatment. Patients with lower levels of education (odds ratio 0.44, 95% CI 0.23, 0.86), higher scores on the personality dimension of neuroticism (odds ratio 0.58, 95% CI 0.36, 0.92) and those with more severe medical illness (odds ratio 0.97, 95% CI 0.95, 0.99) were less likely to recover with either active or placebo treatments. Elderly women (>60 years of age; odds ratio 0.19, 95% CI 0.05, 0.66) were also less likely to respond to all treatments; however, females had a significantly higher response to placebo treatment compared to males. The factors associated with lack of response to treatment included lower-levels of education, high neuroticism, more severe medical illness and being an older female. This analysis is based on patients agreeing to participate in a randomized controlled trial, limiting representativeness of the sample, however, the demographic and clinical characteristics are common in elderly depressed primary care patients, and may signal the need for increased mental health specialty consultation.