Objective. To present results for the first 3 years of the occupational asthma registry of the Surveillance of Work-related and Occupational Respiratory Diseases in South Africa (SORDSA) programme, ending December 1999. Design. Surveillance was accomplished by collecting voluntary reports of occupational asthma cases from pulmonologists, occupational medicine practitioners and occupational health nurses. Setting. Medical and occupational health referral centres in the nine provinces of South Africa. Subjects. patients diagnosed with new-onset occupational asthma with latency or irritant-induced asthma, reported to SORDSA during 1997 - 1999. Outcome measures. Frequencies of cases, causative agents, industries causing exposure and diagnostic methods. Average annual incidence rates by province and by occupation. Results. During this period 324 cases of occupational asthma were reported. The average annual incidence rate of occupational asthma was estimated in the three best-reporting provinces, namely Gauteng, KwaZulu-Natal and the Western Cape, as 17.5 per million employed people annually. This rate was highest in the Western Cape (25.1 per million). Semi-skilled operators had the highest incidence rate of 68.7 per million annually in the three provinces. Isocyanates and latex were the most common agents. Low molecular weight causative agents predominated (68.8%) over high molecular weight agents. Health care was the most frequently reported workplace for occupational asthma (OA) development. Serial peak flow testing was the method most often used for diagnosis. One-fifth of the cases were still occupationally exposed to the causative agent at time of diagnosis. Conclusion. Despite underreporting, SORDSA's estimate of the occupational asthma incidence rate was similar to that of the UK. SORDSA has also demonstrated that surveillance programmes in a developing country can provide useful information on which to base prevention activities.