This paper analyses the relationship between social capital and population health. The analysis is carried out within an econometric model of population health in 19 countries in the Organisation for Economic Co-operation and Development countries using panel data covering three different time periods. Social capital is measured by the proportion of people who say that that they generally trust other people and by membership in voluntary associations. The model performs well in explaining health outcomes. We find very little statistically significant evidence that the standard indicators of social capital have a positive effect on population health. By contrast, per capita income and the proportion of health expenditure financed by the government are both significantly and positively associated with better health outcomes. The paper casts doubt upon the widely accepted hypothesis that social capital has a positive effect on health and illustrates the importance of testing this kind of hypothesis in an extended model. (C) 2002 Elsevier Science Ltd. All rights reserved.