A major constraint to obtaining reliable information about blindness and its causes in developing countries is the limited availability of ophthalmologists for diagnosis in population-based surveys. This study in rural south India assessed the feasibility of using non-ophthalmologists to make diagnoses in a population-based survey. Ten men in their early twenties with 12 years of schooling were recruited and trained as ophthalmic assistants through a six week course in basic ophthalmology. All people aged 40 and older in households in 24 villages were enumerated and invited to have an eye examination. At a central site, 1309 subjects were independently examined by an ophthalmologist and two different ophthalmic assistants. Ophthalmic assistant cataract diagnosis is both sensitive and specific relative to the ophthalmologist's diagnosis. Sensitivity and specificity estimates were used to adjust prevalence estimates obtained from ophthalmic assistant examinations conducted at the central site as well as at the doorstep of sample households. The findings indicate that epidemiologic assessment of cataract blindness can be completed using non-ophthalmologists to diagnose cataract.