ABSTRACTBackground: Cardiorespiratory fitness and body fatness are bothrelated to health, but their interrelation to all-cause and cardiovascu-lar disease (CVD) mortality is unknown.Objective: We examined the health benefits of leanness and the haz-ards of obesity while simultaneously considering cardiorespiratoryfitness.Design: This was an observational cohort study. We followed21925 men, aged 30–83 y, who had a body-composition assess-ment and a maximal treadmill exercise test. There were 428 deaths(144 from CVD, 143 from cancer, and 141 from other causes) inan average of 8 y of follow-up (176742 man-years).Results: After adjustment for age, examination year, cigarettesmoking, alcohol intake, and parental history of ischemic heartdisease, unfit (low cardiorespiratory fitness as determined by max-imal exercise testing), lean men had double the risk of all-causemortality of fit, lean men (relative risk: 2.07; 95% CI: 1.16, 3.69;P =0.01). Unfit,lean men also had a higher risk of all-cause andCVD mortality than did men who were fit and obese. We observedsimilar results for fat and fat-free mass in relation to mortality.Unfit men had a higher risk of all-cause and CVD mortality thandid fit men in all fat and fat-free mass categories. Similarly, unfitmen with low waist girths (< 87 cm) had greater risk of all-causemortality than did fit men with high waist girths (≥ 99 cm).Conclusions: The health benefits of leanness are limited to fitmen, and being fit may reduce the hazards of obesity. Am JClin Nutr 1999;69:373–80.KEY WORDSBody composition, cardiorespiratory fitness, epidemiology,mortality, cardiovascular disease mortality, all-cause mortality,fat mass, fat-free mass, waist girth, menINTRODUCTIONObesity is a public health problem in the United States (1) andthe prevalence of obesity has increased substantially over the pastfew decades (2). However, the health effects of body fatness in rela-tion to longevity are unclear. Many studies show increased mortal-ity in the leanest as well as the most obese individuals (3–6), butothers do not observe this trend (7–9). Manson et al (9) suggest thatfindings of high mortality rates in individuals with low weight-for-height are associated with methodologic limitations such as failureto control for cigarette smoking, failure to eliminate early mortal-ity due to preexisting disease, and inappropriate control for obe-sity-related biological factors. Nonetheless, a recent meta-analysisdocumented elevated mortality in association with leanness afteraccounting for smoking and preexisting disease (10).Another unexplored methodologic limitation in obesity researchis that body mass index (BMI; in kg/m2) is commonly used toexamine the obesity-mortality association even though BMI is notan accurate measure of obesity. Rather, it mainly indicates over-weight for height but does not discriminate between fat mass andfat-free mass (FFM). Some studies show higher death rates in indi-viduals with low BMIs and high waist-to-hip circumference ratios(WHRs), but not in those with high BMIs and low WHRs (11–13).The health effects of overweight on height and body compositionin relation to cardiovascular disease (CVD) risk factors need fur-ther research (14, 15). There has been little research on the relationbetween measured body fatness and mortality (16).We believe that cardiorespiratory fitness should also be consideredin examining the relation between body composition and mortality.Cardiorespiratory fitness is a powerful predictor of all-cause and CVDmortality (17–19) and appeared to attenuate the relation between BMIand mortality in an earlier study (20). However, the health effects ofbody fatness and cardiorespiratory fitness in relation to longevityremain unexplored. Therefore, the purpose of this study was to exam-ine the health consequences of body fatness and cardiorespiratory fit-ness in relation to all-cause and CVD mortality in men. We alsoassessed the associations of fat mass, FFM, and waist circumference tomortality after taking cardiorespiratory fitness into account.SUBJECTS AND METHODSSubjects and measurementsSubjects were 21925 men aged 30–83 y who had completepreventive medical evaluations between 1971 and 1989 at theAm J Clin Nutr 1999;69:373–80. Printed in USA. © 1999 American Society for Clinical NutritionCardiorespiratory fitness, body composition, and all-cause andcardiovascular disease mortality in men1–3Chong Do Lee, Steven N Blair, and Andrew S Jackson3731From the Division of Epidemiology & Clinical Applications, CooperInstitute for Aerobics Research, Dallas, and the Department of Health andHuman Performance, University of Houston.2Supported in part by US Public Health Service research grant AG06945from the National Institute on Aging, Bethesda, MD, and Polar Electro Oy,Kempele, Finland.3Address reprint requests to SN Blair, 12330 Preston Road, Dallas, TX75230. E-mail: [email protected] February 11, 1998.Accepted for publication August 4, 1998.Original Research CommunicationsCooper Clinic in Dallas. All subjects were residents of theUnited States and had no personal history of myocardial infarc-tion, stroke, or cancer at baseline. All received body compositionassessments and reached ≥ 85% of their age-predicted maximalheart rate [220 ! age (in y)] during their treadmill tests.The study protocol was reviewed and approved annually by theInstitutional Review Board. All subjects gave their informed,written consent for the medical evaluation and subsequent regis-tration in the follow-up study. The medical evaluation, performedafter subjects had fasted overnight for ≥ 12 h, included a physicalexamination, anthropometric measurements, electrocardiogram,blood chemistry analyses, blood pressure assessment, maximalexercise treadmill test, self-report of health habits, and recordingof demographic characteristics. Additional details of examinationprocedures are published elsewhere (17–19).Serum samples were analyzed by automated techniques in alaboratory that participates in the Centers for Disease Controland Prevention Lipid Standardization Program, and blood pres-sure was measured by auscultatory methods with a mercurysphygmomanometer. Body weight and stature were measuredwith a standard