Body mass index (BMI) in conjunction with indices of energy turnover, e.g. physical activity levels (PAL), was recently proposed for classifying adult chronic energy deficiency (CED). Three deprived populations in Africa and Asia were chosen to assess the classification system. Repeated monitoring showed that the combined indicator was affected by instability in PALs and produced implausible discontinuities in the prevalence of different grades of CED. The use of BMI alone produced coherent data with changing BMI distributions, and only 5% of the population would have been wrongly classified as being malnourished, because of being thin but active. The risk of misclassification would be even smaller for populations with BMI distribution shifted towards the right. The prevalence of CED was consistently related within each country to indices of socio-economic status. Yet in Zimbabwe 18% of women and 6% of men had Grade I obesity compared with 11% and 14%, respectively, with CED. Less than 1% Indian and Ethiopian adults were obese but 61% of women and 70% of men were classified as CED in India and 57% and 50%, respectively, in Ethiopia. We propose that adult BMI alone is sufficient to provide important new insights into the problems of food availability and its control in less developed countries.