There is a growing recognition that a better understanding of the heterogeneity of motives and determinants of labour market choices is needed to inform policies intended to redress current maldistribution of health workers, particularly in developing countries. This thesis explored the influence of altruism and other individual characteristics on nurses' job preferences in South Africa, to investigate the impact of potential policy interventions designed to attract nurses to under-served areas. Primary data collection was carried out with a sample of 377 nurses. First, measures of altruism were constructed by playing the dictator game, a behavioural economic game. The nurses showed greater altruism than is usually seen in such experiments, suggesting that more altruistic individuals self-select into the nursing profession. Since actual choices could not be observed, a labelled choice experiment was then used to reproduce the job opportunities offered to nurses at the beginning of their career. In the analysis of nurses' job preferences, a positive correlation was found between pro-social values and preferences for public jobs, while individuals from rural backgrounds were more likely to prefer rural jobs. Another choice experiment was used to model the effects of potential policy levers to make rural public jobs more attractive. The analysis showed that packages including monetary incentives were often the most powerful, and it confirmed the preferences of nurses from rural backgrounds. Combining the two choice experiments, a Markov model was constructed to predict the long-term effects of different policies on nurses' distribution in the labour market. Building on these outcomes, a cost-effectiveness model compares the effectiveness of several intervention packages to attract nurses to rural areas. This analysis showed that monetary incentives are never cost-effective, unlike education opportunities and the selection of more individuals who were more likely to prefer rural areas, such as people of rural origins.