AIMS/BACKGROUND: The purpose of the project was to develop a technique for evaluating screening policies in the treatment of insulin dependent diabetic retinopathy. The study was concerned with patients who contracted the disease, aged under 35 years. Simulation was used to describe the progress of a cohort of patients through disease and treatment. METHOD: Data, derived from a literature survey of European and American population and clinic studies, were used to model as closely as possible the development of retinopathy in all its stages, together with the effects of screening and treatment. The model output was validated against published data. RESULTS: The results showed that where screening sensitivities are high, the frequency of screening makes little difference to the years of sight saved, but it does make a difference if screening sensitivities are close to 50%. CONCLUSIONS: Although annual screening is normally desirable, biannual screening could be considered where patient compliance and screening sensitivities are both high.