BACKGROUND: A clinical examination has been shown to stratify patients by risk for diabetic foot ulcer; it is unknown whether primary care providers can reproducibly perform this examination. METHODS: One hundred forty-seven consecutive diabetic patients at six Veterans Affairs Medical Centers received a structured history and physical examination of the feet from each of two primary care providers; 88 of these patients were also examined by a foot care specialist. The examination consisted of a previously validated four-component diabetic foot ulcer risk stratification examination and a vascular disease history and physical examination. RESULTS: Seventy-nine percent of patients were assessed as having elevated risk for development of foot ulcer. Interobserver agreement for risk stratification was moderate (kappa = 0.51; 95% CI 0.40, 0.62). The agreement for individual components of the stratification examination was variable, with kappa statistics ranging from 0.36 to 0.91. Primary care providers had good sensitivity and specificity for most components of the examination (compared with foot care specialist's examination as the criterion standard), but frequently were unable to identify pedal pulses [sensitivity 0.52 (95% CI 0.41, 0.61)] or foot deformity [sensitivity 0.51 (95% CI 0.46, 0.56)]. CONCLUSIONS: A validated risk stratification foot examination for diabetic patients is reproducible and largely accurate when performed by primary care providers.