The Idikan adult mortality study is designed to explore the usefulness of the verbal autopsy methodology in the determination of cause-specific adult mortality. Such data have been largely unavailable in developing countries. Members of a stable urban community (4333 adults) were registered in their family units and followed up every 3 months to ascertain deaths, exits and new entries to the study population. Deaths were investigated by means of a verbal autopsy, which was administered by a trained interviewer to the relative closest to the decedent. Two physicians independently studied the results and assigned a cause of death, and these were then confirmed or resolved by a senior physician colleague. Hospitals where deaths occurred were visited and assignable cause of death was obtained where available and compared with the cause assigned by verbal autopsy. Only 8 of the 60 investigated deaths were confirmed to have occurred in a hospital in the first 2 years of the study. The cause of death assigned by verbal autopsy agreed moderately both between the independent coders as well as with the hospital assigned cause of death using the Kappa statistics for agreement beyond chance. We therefore concluded that verbal autopsies appear moderately useful for adult death statistics (Kappa = 0.23 to 0.1). Accumulation of more hospital deaths is needed, however, to make more definite conclusions about the validity of the technique.