Retinopathy is a common complication of insulin-dependent (IDDM) and non-insulin-dependent (NIDDM) diabetes, but serious visual loss may be prevented or delayed with sufficiently early diagnosis and treatment. Screening for early signs of retinopathy is clearly beneficial for some people, but there is no established consensus about who should be screened, by whom, by what technique and with what frequency, especially for NIDDM. The model described in this paper simulates the development of eye disease in a population of NIDDM patients and the effects of different screening schemes in terms of years of sight saved and the numbers of people prevented from suffering severe visual loss. The initial results indicate that blanket screening of all NIDDM patients may not be effective.