The safety and efficacy of ivermectin in the prevention of blindness from onchocerciasis have been established in many studies that have addressed the drug's effects on the front of the eye. We undertook a study with sufficient statistical power to detect an effect on optic nerve disease (OND), probably the main cause of blindness in the disorder. The trial was based in 34 mesoendemic communities in Kaduna State, Nigeria. Villagers aged 5 years and older were randomly assigned annual dosing with ivermectin or placebo for 3 years. Participants underwent medical and ophthalmological examinations before the first, third, and fourth treatments. 3522 villagers aged 15 and older were re-examined at least once. Skin-snip samples were taken at baseline for calculation of microfilarial load. The outcome measure was development of disc pallor accompanied by objective evidence of deterioration in visual function; 116 subjects (45 ivermectin-treated, 71 placebo-treated) showed such changes during the trial. The incidence rate ratio (ivermectin vs placebo) was 0.90 (95% CI 0.54-1.51) for subjects with loads of 0-10 mf (microfilariae) per mg skin and 0.52 (0.29-0.93) for subjects with more than 10 mf/mg. The incidence rate ratio varied little when account was taken of age, sex, presence of pre-existing disc pallor in one eye, previous use of diethylcarbamazine citrate, or doses of ivermectin or placebo received. There was evidence that ivermectin reduced the incidence of OND in subjects with microfilarial loads above 10 mf/mg but had little effect in those with lower loads. Sustained annual delivery of ivermectin could prevent a substantial proportion of onchocercal blindness in mesoendemic communities.