This paper identifies patterns in child growth monitoring data and relates these to long‐term socio‐economic, environmental and physical characteristics of Zimbabwean districts. The data used are not based on small sample primary surveys, but rather are national data sets with broad spatial and temporal • coverage. Data reduction techniques are applied to the interdependent set of potential causative variables to provide four compound determining factors. The relationship between these four factors and growth monitoring data, standardised to 1991, the mid‐point of the times series, is then analysed using linear regression. Whilst the nutritional status of under‐5s in Zimbabwe is shown to have improved, despite severe droughts, the HIV/AIDS epidemic and cutbacks in public expenditure, the proportion of underweight children is higher in areas where sanitation, water and housing facilities are below average. Districts with a better educated population and greater numbers of wealthy households also have lower percentages of underweight children. Use of secondary data in this way can provide an efficient method to identify the patterns and causes of wide area changes in nutritional data and so complement more localised field surveys.