Cigarette smoking is a potential confounder of the relationship between obesity and mortality, and statistical control for this factor requires careful consideration. Framingham Heart Study subjects were weighed, their stature measured, and cigarette smoking histories obtained at the first biennial examination starting in 1949. Of men under desirable weight (Metropolitan relative weight [MRW], less than 100%), more than 80% were smokers, while only about 55% of the extremely overweight men were cigarette smokers. When age-, smoking-, and MRW-specific mortalities for 26 years of follow-up were calculated in these men, it was found that smokers had higher mortality than nonsmokers but that in the smokers and nonsmokers, minimum mortalities occurred for subjects who were initially in the "desirable weight" group (MRW, 100% to 109%). Among cigarette smokers, lean men (MRW, less than 100%) experienced considerably elevated mortality, often higher than that in all but the most overweight cigarette smokers. These findings suggest that elevated mortality in low-weight American men results from the mortality risks associated with cigarette smoking and demonstrates the need for controlling for cigarette smoking when considering the relationship between relative weight and mortality. Furthermore, the concept of "desirable weight" developed by the Metropolitan Life Insurance Co in 1959 and subsequently distributed in tabular form is validated by this long-term study. Thus, even those men who were near the average weight (about 20% above "desirable weight") showed appreciably elevated mortality. This finding is contrary to the widely held view that moderate overweight carries no increased risk.