Coronary heart disease surveillance studies require monitoring of hospitalized events. Retrospective record reviews and patient interviews during hospitalization are common surveillance methods. This study reports the agreement between these two methods in assessing medical history among 4,230 patients enrolled in the Minnesota Heart Survey Registry. Agreements between methods in determining a patient's history of stroke, myocardial infarction, and hypertension were substantial (kappa coefficients > 0.69). Agreements on acute chest pain (kappa coefficient = 0.39) and ever-smoking status (kappa coefficient = 0.43) were only moderate. In determining medical history, retrospective medical record surveillance appears to be comparable to more direct, yet more expensive, contemporaneous methods.