Depressive symptoms are associated with development of type 2 diabetes, but it is unclear whether type 2 diabetes is a risk factor for elevated depressive symptoms. To examine the bidirectional association between depressive symptoms and type 2 diabetes. Multi-Ethnic Study of Atherosclerosis, a longitudinal, ethnically diverse cohort study of US men and women aged 45 to 84 years enrolled in 2000-2002 and followed up until 2004-2005. Elevated depressive symptoms defined by Center for Epidemiologic Studies Depression Scale (CES-D) score of 16 or higher, use of antidepressant medications, or both. The CES-D score was also modeled continuously. Participants were categorized as normal fasting glucose (< 100 mg/dL), impaired fasting glucose (100-125 mg/dL), or type 2 diabetes (> or = 126 mg/dL or receiving treatment). Analysis 1 included 5201 participants without type 2 diabetes at baseline and estimated the relative hazard of incident type 2 diabetes over 3.2 years for those with and without depressive symptoms. Analysis 2 included 4847 participants without depressive symptoms at baseline and calculated the relative odds of developing depressive symptoms over 3.1 years for those with and without type 2 diabetes. In analysis 1, the incidence rate of type 2 diabetes was 22.0 and 16.6 per 1000 person-years for those with and without elevated depressive symptoms, respectively. The risk of incident type 2 diabetes was 1.10 times higher for each 5-unit increment in CES-D score (95% confidence interval [CI], 1.02-1.19) after adjustment for demographic factors and body mass index. This association persisted following adjustment for metabolic, inflammatory, socioeconomic, or lifestyle factors, although it was no longer statistically significant following adjustment for the latter (relative hazard, 1.08; 95% CI, 0.99-1.19). In analysis 2, the incidence rates of elevated depressive symptoms per 1000-person years were 36.8 for participants with normal fasting glucose; 27.9 for impaired fasting glucose; 31.2 for untreated type 2 diabetes, and 61.9 for treated type 2 diabetes. Compared with normal fasting glucose, the demographic-adjusted odds ratios of developing elevated depressive symptoms were 0.79 (95% CI, 0.63-0.99) for impaired fasting glucose, 0.75 (95% CI, 0.44-1.27) for untreated type 2 diabetes, and 1.54 (95% CI, 1.13-2.09) for treated type 2 diabetes. None of these associations with incident depressive symptoms were materially altered with adjustment for body mass index, socioeconomic and lifestyle factors, and comorbidities. Findings in both analyses were comparable across ethnic groups. A modest association of baseline depressive symptoms with incident type 2 diabetes existed that was partially explained by lifestyle factors. Impaired fasting glucose and untreated type 2 diabetes were inversely associated with incident depressive symptoms, whereas treated type 2 diabetes showed a positive association with depressive symptoms. These associations were not substantively affected by adjustment for potential confounding or mediating factors.