Diarrhoea rates are influenced by weather and climate; transmission can be affected by temperature and rainfall extremes, although few studies have quantified this effect. We undertook a global cross-sectional study of diarrhoea incidence in children under 5, drawing on studies published in the last 50 yr, and assessed the association with climate variables. Log-linear regression was used to quantify any association, controlling for the effects of age, socio-economic conditions and access to improved water and sanitation. We found a negative association between rainfall and diarrhoea rates, with a 4 % increase in diarrhoea incidence (95 % confidence interval, CI: 1-7 %, p = 0.02) for each 10 rum mo-1 decrease in rainfall. Little evidence for association with temperature or climate type was found. Our result for rainfall is consistent with a similar study covering a smaller geographic region. Though biases cannot be excluded, the most likely mechanism is that low rainfall leads to water scarcity, which in turn leads to the use of unprotected water sources and reduces hygiene practices. In the future, greater numbers are expected to experience water scarcity, which may lead to more diarrhoea cases in some locations. This study lends support to programmes for hygiene and water and sanitation coverage, as well as lending support to actions to adapt to and mitigate climate change. © Inter-Research 2007.