Although a growing body of economic work has looked at the role of pro-social motives to explain self-selection into public or not-for-profit sector jobs, in particular in the delivery of social services, no attention has been given to the role of pro-social preferences in the decision to take up posts in rural and isolated areas. Yet there are reasons to believe that such choices involve a degree of self-sacrifice, in particular in developing countries where rural regions typically combine geographic hostile environment, high levels of poverty, low educational opportunities, limited access to basic services and worse career opportunities. On the other hand, as shortage of qualified staff is higher in rural areas, the returns on the presence of a health worker, and the benefits to the populations, are likely to be higher. Using data from a longitudinal study of nurses in South Africa this paper tests this hypothesis by linking experimental measure of pro-social preferences and revealed preferences outcomes. Three measures of pro-social preferences are constructed based on donations made by study participants in a dictator game played at baseline. Job choices are observed three years later for more than 97% of the initial sample. We show that the more dedicated the nurses - measured by their generosity towards patients in the dictator game - the more likely they are to have chosen a rural job. This result is robust to the inclusion of various demographic controls and to different econometric specifications. This finding contributes to the literature on role of pro-social values as an intrinsic motivation factor in labour supply decisions and it has policy implications for the provision of social services in difficult settings. (C) 2014 The Authors. Published by Elsevier B.V.