Objectives. To ascertain the urban/rural distribution of health personnel and the opinions of the medical fraternity in KwaZulu-Natal on compulsory rural service for medical practitioners. Design. Cross-sectional analysis of geographical distribution of health personnel in KwaZulu-Natal based on 1991/92 South African Medical and Dental Council, South African Nursing Council and Pharmacy Council registration data. Opinion survey by administration of a structured questionnaire to a simple, random sample of private practitioners, academic consultants, postgraduate and undergraduate medical students and key informants in senior health service management in KwaZulu-Natal. Results. Peripheral rural areas had health personnel/population ratios higher than or equivalent to those of urban areas, whereas the ratios were 15 - 40 times lower in deep rural areas, The key finding of the opinion survey was that the majority of all sectors except fifth-year medical students felt that rural service should be compulsory, either post-internship, prior to specialisation or prior to,entry into private practice, However, respondents were significantly more likely to agree to rural. service that would not affect them personally, The majority (54 - 87%) of all sectors felt that an option of 'buying out' of rural service should not be permitted, Respondents identified a range of financial, health service, academic, infrastructural and social incentives for rural practice, It is recommended that post-internship rural service be compulsory for a period of 6 months to 1 year provided that academic, health service and infrastructural deficiencies are ameliorated and appropriate financial incentives are provided.