Although the prevalence of obesity, a well-known risk factor for various chronic diseases such as hypertension, type 2 diabetes and dyslipidemia, is rapidly increasing worldwide, the association of obesity with all-cause mortality remains controversial. Many previous epidemiologic studies have demonstrated a U-shaped relationship between obesity and mortality, suggesting that there is an obesity paradox. However, recent large-scale meta-analyses found contradictory results that both overweight and obese subjects exhibited a significant increase in all-cause mortality. This review summarizes the key epidemiologic studies on the association of obesity with mortality and thoroughly examines the causes of the obesity paradox and the precautions needed in the interpretation of this clinical evidence.