In a community-based intervention trial to reduce childhood mortality from pneumonia the intervention area included 58 villages (6176 children aged 0-4 years) and the control area 44 villages (3947 children) in Gadchiroli, India. The interventions included mass education about childhood pneumonia and case-management of pneumonia by paramedics, village health workers, and traditional birth attendants (TBAs) who were trained to recognise childhood pneumonia and treat it with co-trimoxazole. Parents sought treatment, and coverage was 76% without active case-detection efforts. The case-fatality rate among the 612 cases treated by health workers was 0.8%, compared with 13.5% in the control area. After a year of intervention pneumonia-specific childhood mortality was significantly lower in the intervention than in the control area (8.1 vs 17.5 deaths per 1000 children under 5 years); the difference between the areas was greatest in children under 1 year. The differences in infant mortality (89 vs 121 per 1000) and total under-5 mortality (28.5 vs 40.7 per 1000) were highly significant. Mortality from other causes remained similar in the two areas but neonatal mortality due to birth injury and prematurity was significantly lower in the intervention area, presumably owing to the combination of better maternal and neonatal care by the TBAs trained in the project and the availability of treatment for pneumonia. The cost of co-trimoxazole was US $0.025 per child per year ($2.64 per child saved).