Background. The mortality and prevalence of diabetes mellitus (DM) vary across racial/ethnic groups with African Americans/blacks being disproportionately affected. However, it is unclear to what extent such disparities persist after the adjustment for covariates related to race/ethnicity and/or DM in the population. We aimed to assess racial/ethnic disparities in DM and to determine which covariates account for the observed racial/ethnic variabilities. Materials and Methods. We utilized a large cross-sectional survey of the US noninstitutionalized residents (n = 30, 852) to investigate the racial/ethnic disparities in diabetes mellitus, and the degree in which the disparities are explained by the relevant covariates. Pearson's chi-square was used to examine study variables by race/ethnicity, while logistic regression was used to assess the effect of race/ethnicity and other covariates on DM prevalence. Results. There were statistically significant ethnic/racial differences with respect to income, education, marital status, smoking, alcohol, physical activities, body mass index, and age, p < 0.05, but not insurance coverage, p > 0.05. Race/ethnicity was a single independent predictor of DM, with African Americans (non-Hispanic blacks) more likely to be diagnosed for DM compared with non-Hispanic whites, prevalence odds ratio (POR) 1.45, 95% confidence interval (CI) 1.30-1.62, while Hispanics were less likely to be diagnosed, POR = 0.98, 95% CI 0.87-1.09. Similarly, after controlling for potential confounders, the racial/ethnic variability in DM between AA/blacks and non-Hispanic whites persisted, adjusted POR = 1.30, 95% CI 1.15-1.47. Conclusions. Racial/ethnic variability exists in DM prevalence and is unexplained by racial/ethnic variance in education, income, marital status, smoking, alcohol, physical activities, age, and sex.