Previous studies have reported that reduced mortality rates in hemodialysis (HD) patients were negatively related to body mass index (BMI). The potentially protective effect of increased BMI in HD patients has been referred to as 'reverse epidemiology'. Our meta-analysis was conducted to examine the relationship between different BMI ranges and mortality in HD patients. Eligible studies assessing the effects of BMI ranges on all-cause mortality (published from 1966 to February 2012) were searched, using 'hemodialysis' or 'haemodialysis' and 'obese' or 'body mass index' or 'overweight' as key words, in combination with 'mortality', 'survival', 'reverse epidemiology' and 'obesity paradox'. Inclusion criteria were that trials reported mortality in HD patients according to the traditional World Health Organization/National Institutes of Health BMI classification, and BMI levels are acceptable within 2 index points. The quality of the trials was evaluated using the risk of bias assessment in studies included in Cochrane reviews. The mortality rates in HD patients were the primary end point of the study. With no significant heterogeneity, a fixed-effects model was used for analyses. Four studies with a total of 81,423 patients met final inclusion criteria. Compared to individuals with non-elevated BMI, those with elevated BMI (BMI ≥25, OR 0.67, 95% CI 0.65-0.68) had a lower all-cause mortality. In a risk-adjusted sensitivity analysis, elevated BMI levels (adjusted hazard ratio 0.94, 95% CI 0.92-0.96) remained protective against mortality. High BMI levels were associated with lower all-cause mortality rates in HD patients. It is possible that more stable hemodynamic status, cytokine and neurohormonal alternations contribute to the protective effects of BMI on mortality in HD patients. There is a need for prospective studies to elucidate mechanisms behind this relationship. Copyright © 2012 S. Karger AG, Basel.