PURPOSE To describe the research design and eye examination protocol of The National Blindness and Low Vision Prevalence Survey of Bangladesh and to present the main results of the rural pilot study. METHODS A thorough description of the sampling strategy, eye examination protocol and operational definitions are presented. Multi-stage stratified (rural/urban) cluster random sampling, with probability proportional-to-size procedures, will be used for selection of a cross-sectional, nationally representative sample (12,900 subjects) of the population aged > or = l30 years. Each subject will be tested for visual acuity, auto-refracted and undergo optic disc examination. Those with <6/12 visual acuity in either eye will be re-tested with their refractive correction, dilated and examined for anterior and posterior segment disease. A preliminary, separate rural pilot survey sample was conducted according to the eye examination procedures, with results reported herein. RESULTS Two-hundred-and-four (73.1%) of 279 eligible subjects were examined for the rural pilot. Forty-eight persons had presenting visual acuity worse than 6/12 in either eye. The presenting visual acuity of the better eye was used to group the subjects into the following WHO categories (brackets enclose the number of subjects after refractive correction): Blind: 4 ; Visually impaired: 29 ; Sighted: 171 , of whom 3  were unilaterally blind. Cataract was the main cause of visual acuity of less than 6/12, followed by refractive error, and age-related macular degeneration. CONCLUSIONS The pilot survey demonstrated that the proposed examination process for the main survey was both feasible and appropriate for the purposes of this study. Particular strengths of the pilot and subsequent main survey include the use of logMAR visual acuity testing and auto-refraction of all subjects. The pilot study reveals the burden of cataract and refractive error, which are two of the five diseases specifically targeted by the WHO global blindness initiative 'Vision 2020'.