Many studies have shown that carotid intima-media thickness (IMT) is associated with cardiovascular disease (CVD). Although it remains inconclusive whether assessment of carotid IMT is useful as a screening test for CVD in Japanese diabetic patients, a total of 271 patients (151 men aged 66 ± 10 (standard deviation) years and 220 women aged 71 ± 8 years) were divided into two groups based on the presence of CVD. We cross-sectionally assessed the ability of carotid IMT to identify CVD corresponding to treatment that was examined by receiver-operating characteristic (ROC) curve analyses. Among the 271 diabetic patients, 199 non-CVD and 72 CVD patients were examined. Multiple linear regression analysis using the presence of CVD as an objective variable showed that carotid IMT (β = 0.259, P < 0.001) as well as other confounding factors was a significant independent contributing factor. The ROC curve analysis showed that the best marker of CVD was carotid IMT, with an area under the ROC curve of 0.718 (95% confidence interval (CI), 0.650-0.785). The greatest sensitivity and specificity were obtained when the cut-off value of mean carotid IMT was set at 0.95 mm (sensitivity = 0.71, specificity = 0.60, and accuracy = 0.627). Our study suggests that carotid IMT may be useful for screening diabetic patients with CVD.