Background: Traditionally, blindness surveys have modelled themselves on the "gold standard" of a census and examination of a whole population. Blindness, however, is a relatively rare condition even in badly affected communities; hence, large sample sizes are required to gain adequate estimates of prevalence, particularly by cause. Methods: Three assessments of blindness prevalence and aetiology in the some communities are reported. One involved asking individuals questions concerning their visual status during a census (perceived visual status, PVD), one involved examination of all ostensibly visually disabled people presenting to a central point within each community (examination of the visually disabled, EVD) and the final assessment involved a gold standard examination of the whole population (whole community examination, WCE). Results: In a population of 8139 the blindness prevalence was 2.7% PVS, 3.6% EVD, and 3.1% WCE. Attributed causes of blindness were not representative in the PVS except for cataract. The END yielded cause specific estimates not far from those found at WE except for a relative under-representation of glaucoma and optic atrophy. Conclusion: Since cataract is, by a significant margin, the most common cause of blindness in the world such a simple method as asking individuals if they are blind and what they believe to be the cause may yield adequate estimates of the problem for planning eye care strategies for this condition. Alternatively, an ophthalmologist visiting villages and examining allcomers for visual disability may provide reasonably accurate cause specific prevalence estimates without the expense of a major blindness survey.