Greater consumption of red meat has been linked to a higher risk of mortality and chronic diseases, including diabetes. We aim to examine the associations between total, processed and unprocessed red meat intake and diabetes and to evaluate the substitution effects of other protein sources for red meat on diabetes. This population-based cross-sectional study utilised data from the National Health and Nutrition Examination Survey (NHANES) 2003–2016. Diabetes was defined as a self-reported diagnosis by a physician or other health professional, having a fasting plasma glucose of 126 mg/dl or higher, an HbA1c level of 6·5 % or higher, or the use of antidiabetic drugs. Multivariable logistic regression models were conducted. The study included 34 737 adult participants (mean (sd) age of 45·8 (17·5) years) from NHANES 2003–2016. After adjusting for major confounders, compared with the first quintile, higher intakes of total, processed and unprocessed red meat were positively associated with higher odds of diabetes, with adjusted OR of 1·49 (95 % CI 1·22, 1·81), 1·47 (95 % CI 1·17, 1·84) and 1·24 (95 % CI 1·06, 1·44), respectively. The corresponding P-trend values were (< 0.001, 0.001, and 0.006). In this nationally representative sample of US adults, participants in the highest quintiles of total, processed and unprocessed red meat intake had higher odds of diabetes than those in the lowest quintile. Substituting 1 serving/d of dietary protein from foods of plant origin (including nuts, seeds, legumes and soya) for total, processed or unprocessed red meat was associated with 9 % to 14 % lower odds of diabetes.