Health is a key component of the human development index. This paper looks at how health is measured, how the level of health across countries is converging, and which countries are outliers to this global trend. We argue that conceptually health measures should account for illness as well as mortality. However, in practice we show that population mortality and illness measures tend to move closely together, allowing us to use life expectancy as a reasonable proxy for population health. Overall health is improving, and over the last 40 years life expectancy has been converging, with larger gains taking place in countries that initially had lower levels of life expectancy. We show, however, that a detailed analysis gives a more complex picture. Rather than a long term pattern of global convergence we see two distinct groups of countries in the data, clustering around different long run levels of life expectancy. We consider outliers from the general picture found in cross-country analysis. HIV/AIDS plays a large role in explaining the poor health performance of some countries particularly in Sub-Saharan Africa. HIV/AIDS has meant that from 1990 on the process of convergence in health has stopped and is being reversed. Finally we argue that health improvements do not have to wait for national income to rise. Many countries have experienced large health gains without prior income gains, and in countries not affected by HIV/AIDS the last 40 years have largely been a success story in terms of achievements in health.