OBJECTIVE: We conducted comprehensive assessments of emotional distress to examine relations with diabetes medication adherence over time. RESEARCH DESIGN AND METHODS: Ethnically and socioeconomically diverse adults treated for type 2 diabetes completed validated self-reports (SRs) for diabetes distress and depression, were administered semistructured depression interviews, and provided blood samples for A1C. Medication adherence among 104 participants was electronically monitored (EM) over the subsequent 3 months; validated SRs of medication adherence were also obtained. Hierarchical linear regression evaluated independent effects of diabetes distress and depression on adherence. RESULTS: Mean ± SD 3-month medication adherence was 76.1% ± 25.7% for EM and 83.7% ± 21.9% for SR. Higher levels of SR (P < 0.001) and interview-based (P < 0.05) depressive symptom severity (P < 0.05) and diabetes-related distress (P < 0.01) showed a significant bivariate association with EM and SR nonadherence. Regression models showed baseline diabetes distress was a significant independent predictor of EM (β = -0.29; P = 0.001) and SR adherence (β = -0.24; P < 0.02) at follow-up. SR depression was an independent predictor of EM and SR adherence and reduced the effects of diabetes distress to nonsignificance. Subsequent models indicated this effect was driven by somatic rather than cognitive-affective symptoms of depression. Results were consistent but weaker for interview-based depressive symptoms. CONCLUSIONS: Findings support diabetes-related distress and depression symptom severity as risk factors for type 2 diabetes medication nonadherence. Somatic symptoms captured by depression measures, but not cognitive-affective symptoms, independently predict nonadherence and should be further investigated as a potential link between emotional distress and nonadherence.