Ruminative responses to depression have predicted duration and severity of depressive symptoms. The authors examined how response styles change over the course of treatment for depression and as a function of type of treatment. They also examined the ability of response styles to predict treatment outcome and status at follow-up. Primary care patients (n = 96) with dysthymia or minor depression were randomly assigned to problem-solving therapy, paroxetine, or placebo. Patients' depressive symptoms and rumination, but not distraction, decreased over time. Pretreatment rumination and distraction were associated with more depressive symptoms at the conclusion of treatment; the latter finding was not consistent with the response style theory of depression. Results are discussed in terms of their implications for this theory.