Background: This study was designed to test the hypothesis that diabetes patients with a depression diagnosis are less likely than those without a depression diagnosis to receive recommended frequencies of glycated hemoglobin (A1c) and lipid testing. Methods: We identified 14,144 adults with diabetes receiving care from a single large multispecialty medical group in Minnesota between 1997 and 2003. Those with two or more depression ICD-9 codes, or with one depression ICD-9 code plus at least one prescription fill for depression medication were classified as having depression. A total of 2179 met these criteria. Subjects were tracked for 4 years. Data on A1c and lipid tests were extracted from claims data and verified by chart audit in a sub-sample of patients. Results: The 2179 individuals with at least one depression episode had a mean age of 60 and had 8.2 primary care visits, on average. Eight hundred thirty-five (38%) of these individuals were male, 1886 (86%) took an antidepressant, 570 (26%) had at least one other comorbid condition, 642 (29%) disenrolled in the plan prior to 20 months, and 259 (12%) were newly diagnosed cases of diabetes. Relative to those without a depression diagnosis, those with a depression diagnosis had significantly fewer (P<0.001) than recommended A1c and low density lipoprotein (LDL) tests in bivariate analysis (30% compared with 35% among non-depressed cases). This finding persisted after adjustment in multivariate models for age, sex, comorbidity, number of primary care visits, and duration of the diabetes episode. Conclusions: Episodic depression is associated with sub-optimal monitoring of A1c and LDL levels in diabetes patients. These findings may help direct efforts to improve diabetes care in those with depression. They also suggest accountability measures used to assess quality of diabetes care may need to be adjusted for differences in depression occurrence across providers, clinics, or medical groups.