The study was carried out in a rural population in central India. A random sample of 11 village communities provided 1020 persons aged 40-64 years, who were examined in 1982 and again reassessed in 1986. Statistical analysis, based on the Mantel-Haenszel method for stratified data, showed increased mortality in persons who had central lens opacities, compared with those who had trivial or no central lens opacities. The significant age-adjusted death ratio was just over 2 (2.2), as were the age/sex-adjusted and age/vision-adjusted estimates, which indicate doubling of mortality in the cataract cohort. Multiple regression analysis using the Cox proportional-hazards model gave very similar results. Statistical tests for homogeneity of death ratios across the various age/sex/vision strata were carried out, and the observed association between cataract and mortality was found to be consistent, both in males and in females, in the youngest and oldest age groups, and among those with adequate vision of 6/18 or better as well as among persons with serious visual impairment. There were no known diabetics in the study sample, which came from what could reasonably be regarded as a non-diabetic population.