Background/Aims: Depression is a pervasive and costly medical issue, but little is known about predictors of poor response to the usual care of depression in primary care. This study examines what patient characteristics best predict non-response to usual care for depression in primary care. Methods: Design: Observational Setting: 78 primary care clinics across Minnesota. Participants: 537 primary care patients treated with antidepressants. Intervention: Usual care for depression. Outcome: Response and remission from depression accessed via PHQ-9 scores at 6-months after baseline. Results: The majority of patients seen for usual care of depression were female, Caucasian, and employed; most had some college education and rated their health as good-to-excellent. At baseline, 59% had mild depression, 30% moderate, and 11% severe as assessed by the PHQ-9. Less than a third of patients had psychotherapy or psychiatric referral in addition to their antidepressant medication. At six months only 46% of patients obtained response or remission from depression. Patients were significantly less likely to respond to usual care for depression if they were unemployed (OR=0.6, 95%CI: 0.4-0.8), had lower household incomes (OR=0.6, 95%CI:0.4-0.9), had poor-to-fair selfreported physical health (OR=0.5, 95%CI:0.3-0.7), or felt their life was greater than 50% impaired due to health issues (OR=0.7, 95%CI:0.5-1.0). Neither depression severity nor recurrent episode was a significant predictor of response. Conclusions: Providers and care systems may be able to improve patient outcomes for depression by focusing more intensive interventions on those least likely to respond to usual care: patients who are unemployed or have lower incomes, poorer self-rated health, or life impairment due to health issues.