Mseleni joint disease (MJD), an endemic osteoarthrosis, was first described in 1970, localized to Kwa Zulu, South Africa. In 1981 another survey was conducted (1) to see whether prevalence had changed, (2) to estimate the prevalence in areas not previously surveyed and (3) to estimate its socioeconomic impact. A 5% stratified random sample of homesteads was selected and five local interviewers surveyed 333 kraals, interviewing 333 individuals, obtaining information on 3368 live members of these homesteads. Overall prevalence rates were 7.4% in 1566 females and 3.0% in 1179 males (age or sex unknown in 623). The previously affected area still had the highest prevalence rates, but they were significantly lower than before, even if only residents of at least ten years' duration were considered. These consistently lower age-specific prevalence rates are compatible with a decline in incidence rate, possibly because of disappearance of an environmental causal agent. Prevalence rates for females in some areas not previously surveyed were between 5 and 10%, suggesting that a larger area than stated before is affected. The proportions of severely disabled individuals are the same as reported in 1973 but greater in number considering the larger area affected. Of these, an estimated 67% of males and 47% of females are not receiving disability grants. Fifty per cent of children (aged 6-25 years) of affected parents have had no schooling compared with 30% of children of unaffected parents. This study also highlighted several methodological problems unique to rural research. It is suggested that while aetiological research should continue, the existing prevalence of MJD has major social, economic and health care implications, for which solutions should be researched as a high priority.