BACKGROUND/AIMS: Obesity is a predisposing factor for chronic illnesses such as type 2 diabetes, heart disease and cancer. In chronic kidney disease (CKD), the effect of obesity on mortality is reversed. Obese patients appear protected. Two ideas have been advanced to explain this 'reverse epidemiology'. First, obesity may buffer patients from wasting. Second, fat may sequester uraemic toxins leading to a systematic error in the prescription of dialysis. Our aim was to use data on the scaling of daily energy expenditure, fat and lean tissue mass to predict the pattern of variation in mortality with obesity under the contrasting hypotheses. METHODS: We used data on daily energy demands measured using the doubly labelled water technique and body composition collected on a cohort of 503 individuals to model the expected impacts of wasting and fat sequestration/underdialysis on mortality. RESULTS: A model predicting mortality due to wasting replicated the mortality pattern of the obesity paradox. However, quantitatively the beneficial effect of being fat was predicted to be much larger than that observed in the actual CKD population. Similar results were found for the fat sequestration/underdialysis hypothesis, but in this case the discrepancy was smaller. DISCUSSION: These models tend to support the fat sequestration and underdialysis idea more than the wasting hypothesis. In part (or in whole) this may be because of inadequacies in the model construction which are currently based on rather crude assumptions. Refinement of the models may enable better tests between alternative ideas for the obesity paradox.