Journal Logo


Exercise Attenuates the Association of Body Weight with Diet in 106,737 Runners


Author Information
Medicine & Science in Sports & Exercise: November 2011 - Volume 43 - Issue 11 - p 2120-2126
doi: 10.1249/MSS.0b013e31821cd128
  • Free


Excess body weight is the accumulated effects of small positive imbalances between energy consumption and expenditure over time (9). However, these imbalances are small and difficult to detect in population studies (9). Population studies may have greater success in identifying factors affecting energy balance rather than documenting the energy balance itself. In this regard, the strongest evidence seems to relate weight gain from higher meat and lower fruit, vegetable, and starchy food intakes (3,10,26,27,29). These foods may be directly causative, or they may be indicative of high- and low-energy density diets.

Most epidemiological studies estimate energy expenditure from the sum of different physical activities during a 24-h day or during recreation (2). These studies were recently interpreted as generally failing to show that physical activity affects weight gain prospectively (29). In contrast, we have published strong and consistent dose-response relationships between runners' usual distance run and prospective weight gain (34,39). Specifically, we have shown that running attenuates age-related weight gain prospectively in proportion to the exercise dose (34) and that increasing and decreasing exercise produces reciprocal changes in body weight (39). Our data also suggest that an exercise hiatus may lead to a net weight gain even when exercise resumes at former levels (35). Our greater success in identifying these relationships may relate to running being a vigorous physical activity (i.e., expending more than sixfold the energy expenditure of being at rest), the energy expenditure of which is more simply based on total distance run rather than the more complicated sum product of perceived time and intensity (2).

In 1997, we reported that body mass index (BMI) increased in association with greater meat and lesser fruit consumption and that these relationships seemed to be attenuated by vigorous physical activity (31). This initial observation, on the basis of only 7054 men and 1837 women, did not provide the statistical power required to test whether the attenuation with physical activity was statistically significant. The current article tests whether exercise can attenuate the effects of diet on total and regional adiposity in a very large sample of more than 100,000 runners. The magnitudes of the interactions and their extreme statistical significance rule out any reasonable probability of their being due to chance. The attenuating effect of exercise for preventing weight gain due to high-risk diets is not currently recognized and may provide an important intervention tool for combating obesity in the "obesogenic" environment.


A two-page mailed questionnaire, sent to subscribers of a running magazine and to participants of running events, solicited information on demographics (age, race, and education), running history (age when began running at least 12 miles·wk−1, current average weekly mileage, and number of marathons during the preceding 5 yr and best marathon and 10-km times), weight history (greatest and current weight; weight when started running; least weight as a runner; body circumferences of the chest, waist, and hips; and bra cup size), diet (vegetarianism and the current weekly intakes of alcohol, meat, fish, fruit, vitamins C and E, and aspirin), current and past cigarette use, history of heart attacks and cancer, and medications for blood pressure, thyroid, cholesterol, or diabetes. Running distances were reported in miles per week, body circumferences were reported in inches, and body weights were reported in pounds. These values were converted to kilometers per day, centimeters, and kilograms, respectively. The test-retest correlations for self-reported distance run per week (r = 0.89) (33) compare favorably with physical activity scores reported by others.

Intakes of meat, fish, and fruit were based on the questions, "During an average week, how many servings of beef, lamb, or pork do you eat?" and "During an average week, how many pieces of fruit do you eat?" Alcohol intake was estimated from the corresponding questions for 4-oz (112 mL) glasses of wine, 12-oz (336 mL) bottles of beer, and mixed drinks and liqueurs. Alcohol was computed as 10.8 g per 4-oz glass of wine, 13.2 g per 12 oz bottle of beer, and 15.1 g per mixed drink. Correlations between these responses and values obtained from 4-d diet records in 110 men were r = 0.46 and r = 0.38 for consumptions of meat and fruit, respectively. These values agree favorably with published correlations between food records and more extensive food frequency questionnaires for red meat (r = 0.50) and somewhat less favorably for fruit intake (r = 0.50) (8).

The runners' BMI were calculated as the weight in kilograms divided by height in meters squared. Self-reported body circumferences of the waist, hip, and chest were in response to the statement "Please provide, to the best of your ability, your body circumferences in inches" without further instruction. The relationships between circumferences and running distance are expected to be weakened by different locations of where waist, hip, and chest circumferences were measured. However, unless the perceived location varied systematically in relation to running distance, the subjectivity would be unlikely to produce the relationships reported in the tables and figures. Self-reported height and weight from the questionnaire have been found previously to correlate strongly with their clinic measurements (r = 0.96 for both) (33). Self-reported waist circumferences are somewhat less precise as indicated by their correlations with self-reported circumferences on a second questionnaire (r = 0.84) and with their clinic measurements (r = 0.68) (33). Chest circumference is a measure of upper body obesity that exhibits relationships to plasma leptin levels that are not apparent for waist or hip measurements (28). Thoracic fat has also been related to LDL levels (23). The study protocol was reviewed by the University of California Berkeley Committee for the Protection of Human Subjects, and all subjects provided a signed statement of informed consent.

Statistical analyses.

Results are presented as mean ± SE or slopes ± SE except where noted. With the exception of the sample description of Table 1, all analyses were adjusted for age (age and age squared), and education. Multiple regression analyses were used to test whether reported meat and fruit consumption affected the runners' BMI and body circumferences. Specifically, we tested whether the coefficient for a diet × distance interaction differed significantly from zero in a model that also included their separate main effects. In these analyses, diet was defined as meat intake alone, fruit intake alone, and the combination of meat and fruit intakes. Specifically, reported meat and fruit intakes were included as a single index value using their linear regression coefficient for the least active running category (i.e., within the <2-km·d−1 category). We also divided the sample into running increments of <2, 2-4, 4-6, 6-8, and ≥8 km·d−1 and calculated the regression coefficients for diet (meat, fruit, or best linear combination) separately within each stratum.

Characteristics of runners by reported distance run per day.

We also tested whether the slope of the BMI-versus-diet regression line decreased progressively with longer running distances. In these analyses, we applied a single regression slope for diet to all subjects and then tested the significance of adding a separate slope for runners who ran <2 km·d−1 (its significance implying that the slope for running ≥2 km·d−1 was significantly less than <2 km·d−1). The analyses were then repeated including separate coefficients for both <2 km·d−1 and 2-4 km·d−1 (the significance of the 2- to 4-km·d−1 coefficient implying that the slope for running ≥4 km·d−1 was significantly less than 2-4 km·d−1) and separate coefficients for <2 km·d−1, 2-4 km·d−1, and 4-6 km·d−1 (the significance of the 4- to 6-km·d−1 coefficient implying that the slope for running ≥6 km·d−1 was significantly less than 4-6 km·d−1).


There were 106,737 subjects from the National Runners' Health Survey who provided completed data on height, weight, education, running distance, and intakes of meat and fruit and who did not smoke. Table 1 displays their sample characteristics by distance run. The higher mileage runners tended to be younger and slightly more educated, ate less meat and more fruit, and, if male, drank less alcohol. They were also leaner as measured by BMI and body circumferences. Fourteen and seven-tenths percent (14.7%) of the men reported consuming 0 servings of meat per day, 55.0% reported 0.1-0.5 serving per day, 24.5% reported 0.51-1.0 serving per day, and 5.9% reported >1 serving per day. The corresponding percentages for women were 31.8%, 53.3%, 13.2%, and 1.7%, respectively. Average daily fruit consumption for men and women, respectively, were reported as follows: 2.7% and 2.0% reported 0 intake, 42.1% and 41.4% reported 0.1-1 piece, 31.5% and 34.4% reported 1.1-2 pieces, 16.5% and 16.3% reported 2.1-3 pieces, and 7.1% and 6.0% reported >3 pieces per day.

Associations with reported intakes of meat and fruit in the least active runners.

Table 2 presents regression slopes of BMI and body circumferences versus daily servings of meat and fruit by running distance category. The least active category ran <2 km·d−1. Within this group, the men's BMI and waist circumference increased significantly in association with both higher meat intake and lower fruit intake. The women's BMI and circumferences of the waist, hip, and chest also increased significantly with higher meat intake, and their BMI increased in association with lower fruit intake. The multivariate analyses of Table 3 include both foods simultaneously in the analyses and show that meat and fruit contributed independently to BMI and body circumferences in these low-mileage runners. In fact, their coefficients differed little from their separate regression analyses of Table 2.

Regression slopes for body mass index and circumferences versus reported intakes of meat, in kilograms per meter squared or centimeters per servings per day, and fruit, in kilograms per meter squared or centimeters per pieces per day, adjusted for age and education and stratified by running distance.
Multivariate regression analyses to determine the linear combinations of reported intakes of meat and fruit that best predict BMI and circumferences in runners who averaged <2 km·d−1.

Attenuation of diet-weight relationships at higher activity levels.

Table 2 displays the regression slopes relating diet to body size at different activity levels and the significance of the interaction between distance run and diet on body size. The analyses suggest that meat had a significantly weaker relationship to BMI when running ≥8 km·d−1 than <2 km·d−1 in both men and women (43% and 55% reduction, respectively). Fruit intake was also more weakly related to BMI when running ≥8 km·d−1 vis-à-vis <2 km·d−1 in men (75% reduction) and women (94% reduction). Distance run also significantly affected the relationship of both meat and fruit to waist circumference. The relationships of men's chest circumferences to high meat and low fruit intakes were also significantly attenuated by running mileage, as were the relationships of women's chest and hip circumferences to high meat intake.

The analyses of Table 2 do not show whether higher BMI and larger body circumferences were directly related to high meat and low fruit intakes or whether meat content and fruit content are simply indicators of diets that increase the risk for weight gain. Assuming the latter, the linear combinations of Table 3 provide the best predictors of BMI and body circumferences and serve as indicators of high-risk diets. For example, Table 3 shows that the best predictor of BMI was "0.73 × meat − 0.27 × fruit" in men and "1.28 × meat − 0.13 × fruit" in women. Separate linear combinations were calculated for male and female runners and for BMI and each body circumference. These define the dietary indices for the analyses to follow.

The indices were used to produce the bar graph on Figure 1, which shows the attenuating effects of running distances on the diet-weight relationships. The analyses are the same as those presented in Table 2, except that the indices of high-risk diets replace meat and fruit. The coefficient (slope) for the <2 km·d−1 running category is always one because it represents the subset of runners used to create the index. Coefficients (slopes) less than one measure the percent attenuation due to exercise, i.e., the degree to which exercise reduces the effect of the high-risk diet on BMI and body circumferences. For example, the relationship between the men's diet and BMI was given by the slope 1.0(0.73 × meat − 0.27 × fruit) for <2 km·d−1, 0.68(0.73 × meat − 0.27 × fruit) for 2-4 km·d−1, 0.56(0.73 × meat − 0.27 × fruit) for 4-6 km·d, 0.41 × (0.73 × meat − 0.27 × fruit) for 6-8 km·d−1, and 0.35 × (0.73 × meat − 0.27 × fruit) for ≥8 km·d−1. Thus, relative to the men who ran <2 km·d−1, the relationship of the high-risk diet to BMI was reduced by 32% for runners who ran 2-4 km·d−1, 44% for those who ran 4-6 km·d−1, 59% for those who ran 6-8 km·d−1, and 65% for those ran or exceeded 8 km·d−1. This represented a highly significant decline in the relationship of diet to BMI with increasing exercise (P < 10−15). Comparable results were obtained in women.

Bar chart of the regression slope for BMI versus diet index at different running distances. Significance levels represent the significance of the slope within the range of running distances.

More detailed comparisons of the bar graphs of Figure 1 (not displayed) showed that the effect of diet on BMI was significantly less for male and female runners who ran >2 km·d−1 compared with those who ran ≤2 km·d−1 (P < 10−6 and P < 10−10, respectively), for those who ran >4 km·d−1 compared with those who ran 2-4 km·d−1 (P = 0.0002 and P = 0.0007, respectively), and for those who ran > 6 km·d−1 when compared with those who ran 4-6 km·d−1 (P = 0.0003 in men only). Thus, there are statistically significant incremental reductions in the effect of diet on BMI with increasing running distance through at least 6 km·d−1 in men and 4 km·d−1 in women.

Other analyses suggest that exercise attenuates the effect of diet on regional adiposity in a dose-dependent manner for both male and female runners. The effect of diet on runners' waist circumferences decreased significantly with running mileage for both men (P < 10−15) and women (P = 0.0006). More detailed comparisons showed that the effect of diet on waist circumferences was significantly less in men and women who ran ≥2 km·d−1 than in those who ran < 2 km·d−1 (P < 10−5 and P = 0.02, respectively) and in men who ran ≥4 km·d−1 than in those who ran 2-4 km·d−1 (P < 0.0001). The effect of diet on chest circumference was also attenuated by exercise (P = 0.0001 for both sexes), with the effect being significantly weaker in men and women who ran ≥2 km·d−1 than in those who ran <2 km·d−1 (males: P = 0.05, females: P = 0.0005) and weaker in males who ran ≥4 km·d−1 than in those who ran 2-4 km·d−1 (P = 0.05). The effect of diet on women's hip circumference also declined with increasing mileage (diet × exercise interaction: P = 10−6) such that women who ran ≥2 km·d−1 were less strongly affected than those who ran <2 km·d−1 (P = 0.002).


Because these analyses are cross-sectional, it is possible that self-selection led to the observed associations of Figure 1. These analyses were therefore repeated when adjusted for the runners' preexercise BMI (i.e., BMI when they first started running ≥12 miles·wk−1). The analyses suggest that self-selection did not account for the observed association. Specifically, the effect of diet on current BMI (P < 10−15 for both sexes) and the attenuating effect of exercise (males: P < 10−8, females: P < 10−8) remained significant when adjusted for their preexercise BMI.


Consistent with prospective epidemiological data (3,10,26,27,29), the associations of Table 2 show that reported average meat intake was positively associated with both BMI and waist circumference. The significant association was, in fact, replicated in 10 separate subsets (i.e., males who ran <2 km·d−1, males who ran 2-4 km·d−1,…, females who ran ≥8 km·d−1). Fruit intake was also inversely associated with men's BMI and achieved statistical significance in five separate subsets of the men's data. The epidemiological evidence relating adiposity to dietary intake is weaker for fruit than for meat (22,26,27,29), which explains, in part, the more variable association we observed between fruit and BMI (Table 3). Our limited dietary assessment precluded our being able to identify meat or fruit as the specific food responsible, and therefore, we also considered their combined effects in an index that may reflect high versus low energy-dense food, or greater fast-food or restaurant-prepared than home-prepared food consumption. The primary dietary determinant of body fat has been ascribed to the diet's energy density (1), although others suggest the evidence for this is inconclusive (29). Because these data are cross-sectional, it is not possible to prove a causal relationship between the diet and BMI or body circumferences from our data. Our survey questionnaire did not ask about intakes of vegetables or other foods that may also have contributed to total and regional adiposity.

Our analyses provide statistical confirmation of our initial report that there was a trend for vigorous physical activity to attenuate the association of BMI with higher meat and lower fruit consumption in a dose-dependent manner (31). They extend these initial results by showing that vigorous exercise also attenuates 1) the concordance between men's waist circumference and meat intake, 2) the concordance of the men's chest circumference with higher meat intake and lower fruit intake, and 3) the relationship between women's adiposity (BMI and waist, hip, and chest circumferences) and diet (Table 2).

The diminished effect of dietary composition on the BMI of higher mileage runners could be due to improved fat oxidation with exercise. Defects in whole body and skeletal muscle fatty acid oxidation (11) may predispose individuals to gain weight in an obesogenic environment (12). The defect involves increased reliance on CHO vis-à-vis fat oxidation, which are associated with gains in total weight and fat mass (40). Manifestation of the defect includes a high respiratory quotient (RQ), which is characteristic of obese and postobese individuals and is shown to predict weight gain prospectively (4,17,40). In obese individuals, the RQ has also been associated with low muscle lipoprotein lipase activity (5,40). Lower postprandial uptake of free fatty acids by leg skeletal muscle has been shown to correlate with greater visceral fat mass (4). The impaired ability of obese vis-à-vis lean individuals to oxidize fat may be exacerbated by diet, with obese individuals 1responding with a higher CHO and lower fat oxidation after high CHO consumption, and an inability to increase fat oxidation after high fat consumption (30), a condition described as metabolic inflexibility (7). Plasma triglyceride concentrations, elevated in association with obesity, may be a marker for impaired fat oxidation (24). Vigorous physical activity significantly increases the ability to oxidize fat (25), increases skeletal muscle lipoprotein lipase activity (21), reduces the RQ (25), and reduces plasma triglyceride concentrations (32).

The diminished effect of dietary composition on the BMI of higher mileage runners could also be due, in part, to improved coupling between energy intake and expenditure, such that episodic intakes of energy-dense foods are balanced by reduced energy intake at other times (18). In sedentary individuals, reduced physical activity is not necessarily associated with reduced energy intake, whereas in more active individuals, energy intake and expenditure tend to be correlated (20). At high physical activity levels, the coupling between expenditure and intake can be described as tight (14), with active males reported to be able to adjust future food consumption to drinks of disguised high- and low-energy content (13) and to low- and high-energy preloading (16,19). The improved coupling may be the result of the acute regulation from satiety hormones released by the intestinal tract or longer term effects related to leptin, insulin, and postabsorptive signals associated with macronutrient oxidation (18).


The cross-sectional nature of the current analyses prevents our drawing causal inference between running distance, diet, and adiposity. Although it is possible that lean individuals may self-select to run longer distances, prior analyses of these runners suggest that self-selection accounts for only 25% of the association between running distance and BMI in men and 58% in women (36). Prospective analyses have also shown that their leanness is partially due to exercise-induced weight loss (39) and the attenuation of age-related weight gain in accordance to their exercise dose (34). Our inability to attribute the observed associations to the runners' preexercise BMI also argues against self-selection. We also acknowledge that BMI may be a poor indicator of adiposity in high-mileage runners (15). We also caution that the runners may represent a unique group of individuals who may not be representative of the general population; however, we believe that the basic biological processes relating exercise, diet, and body weight are likely to be shared by all individuals. The majority of the men and women in these analyses fell within the healthy weight category as defined by 18.5 ≤ BMI ≤ 25 kg·m−2; nevertheless, the association described herein is of public health relevance because even within the healthy weight range, greater BMI is associated with a greater incidence of diabetes (37), hypertension (37), hypercholesterolemia (38), and CHD (37).

In conclusion, these observations suggest an important advantage to achieving and exceeding minimum guideline physical activity levels, which is to reduce the risk of weight gain by high-risk diets, as characterized by high meat and low fruit content. The mean energy expenditure of the men and women in our least active exercise category (<2 km·d−1) averaged 398 ± 4 and 392 ± 4 MET·min·wk−1, respectively (calculated as 61.2 MET·min·km−1 [34]), which falls below the minimum physical activity level of 450-750 MET·min·wk−1 recommended by the American Heart Association and the American College of Sports Medicine (6). They may also contradict the notion that the effect of physical activity on weight gain is limited to energy expenditure and the thermodynamics of energy balance and may include the additional effects of improved fat oxidation and appetite regulation.

This research was supported by grant HL094717 from the National Heart, Lung, and Blood Institute and grant AG032004 from the Institute of Aging and was conducted at the Ernest Orlando Lawrence Berkeley National Laboratory (Department of Energy DE-AC03-76SF00098 to the University of California).

There is no industrial relationship to report. P. T. Williams was responsible for all aspects of the study.

The author reports no conflict of interest.

The author thanks Ms. Kathryn Hoffman for her help in collecting the data and reviewing the article.

The results of the present study do not constitute endorsement by the American College of Sports Medicine.


1. Adams T, Rini A. Predicting 1-year change in body mass index among college students. J Am Coll Health. 2007;55:361-5.
2. Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc. 2000;32(9 suppl):S498-504.
3. Bes-Rastrollo M, Sánchez-Villegas A, Gómez-Gracia E, Martínez JA, Pajares RM, Martínez-González MA. Predictors of weight gain in a Mediterranean cohort: the Seguimiento Universidad de Navarra Study 1. Am J Clin Nutr. 2006;83:362-70.
4. Colberg SR, Simoneau JA, Thaete FL, Kelley DE. Skeletal muscle utilization of free fatty acids in women with visceral obesity. J Clin Invest. 1995;95:1846-53.
5. Ferraro RT, Eckel RH, Larson DE, et al. Relationship between skeletal muscle lipoprotein lipase activity and 24-hour macronutrient oxidation. J Clin Invest. 1993;92:441-5.
6. Haskell WL, Lee IM, Pate RR, et al. Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association. Circulation. 2007;116:1081-93.
7. Horton TJ, Drougas H, Brachey A, Reed GW, Peters JC, Hill JO. Fat and carbohydrate overfeeding in humans: different effects on energy storage. Am J Clin Nutr. 1995;62:19-29.
8. Hu FB, Rimm E, Smith-Warner SA, et al. Reproducibility and validity of dietary patterns assessed with a food-frequency questionnaire. Am J Clin Nutr. 1999;69:243-9.
9. Institute of Medicine. Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein, and Amino Acids (Macronutrients). Washington (DC): The National Academies Press; 2005. p. 880-935.
10. Kahn HS, Tatham LM, Rodriguez C, Calle EE, Thun MJ, Heath CW Jr. Stable behaviors associated with adults' 10-year change in body mass index and likelihood of gain at the waist. Am J Public Health. 1997;87:747-54.
11. Kelley DE, Goodpaster B, Wing RR, Simoneau JA. Skeletal muscle fatty acid metabolism in association with insulin resistance, obesity, and weight loss. Am J Physiol. 1999;277(6 Pt 1):E1130-41.
12. Kim JY, Hickner RC, Cortright RL, Dohm GL, Houmard JA. Lipid oxidation is reduced in obese human skeletal muscle. Am J Physiol Endocrinol Metab. 2000;279:E1039-44.
13. King NA, Appleton K, Rogers PJ, Blundell JE. Effects of sweetness and energy in drinks on food intake following exercise. Physiol Behav. 1999;66:375-9.
14. King NA, Tremblay A, Blundell JE. Effects of exercise on appetite control: implications for energy balance. Med Sci Sports Exerc. 1997;29(8):1076-89.
15. Leal-Cerro A, Garcia-Luna PP, Astorga R, et al. Serum leptin levels in male marathon athletes before and after the marathon run. J Clin Endocrinol Metab. 1998;83:2376-9.
16. Long SJ, Hart K, Morgan LM. The ability of habitual exercise to influence appetite and food intake in response to high- and low-energy preloads in man. Br J Nutr. 2002;87:517-23.
17. Marra M, Scalfi L, Contaldo F, Pasanisi F. Fasting respiratory quotient as a predictor of long-term weight changes in non-obese women. Ann Nutr Metab. 2004;48:189-92.
18. Martins C, Robertson MD, Morgan LM. Effects of exercise and restrained eating behaviour on appetite control. Proc Nutr Soc. 2008;67:28-41.
19. Martins C, Truby H, Morgan LM. Short-term appetite control in response to a 6-week exercise programme in sedentary volunteers. Br J Nutr. 2007;98:834-42.
20. Miller WC, Koceja DM, Hamilton EJ. A meta-analysis of the past 25 years of weight loss research using diet, exercise or diet plus exercise intervention. Int J Obes Relat Metab Disord. 1997;21:941-7.
21. Nikkilä EA, Taskinen MR, Rehunen S, Härkönen M. Lipoprotein lipase activity in adipose tissue and skeletal muscle of runners: relation to serum lipoproteins. Metabolism. 1978;27:1661-7.
22. Parker DR, Gonzalez S, Derby CA, Gans KM, Lasater TM, Carleton RA. Dietary factors in relation to weight change among men and women from two southeastern New England communities. Int J Obes Relat Metab Disord. 1997;21:103-9.
23. Pouliot MC, Després JP, Moorjani S, et al. Computed tomography-measured trunk fat and plasma lipoprotein levels in nonobese women. Metabolism. 1989;38:1244-50.
24. Roberts R, Bickerton AS, Fielding BA, et al. Reduced oxidation of dietary fat after a short term high- carbohydrate diet. Am J Clin Nutr. 2008;87:824-31.
25. Romijn JA, Klein S, Coyle EF, Sidossis LS, Wolfe RR. Strenuous endurance training increases lipolysis and triglyceride-fatty acid cycling at rest. J Appl Physiol. 1993;75:108-13.
26. Sánchez-Villegas A, Bes-Rastrollo M, Martinez-González MA, Serra-Majem L. Adherence to a Mediterranean dietary pattern and weight gain in a follow-up study: the SUN cohort. Int J Obes (Lond). 2006;30:350-8.
27. Schulz M, Kroke A, Liese AD, Hoffmann K, Bergmann MM, Boeing H. Food groups as predictors for short-term weight changes in men and women of the EPIC-Potsdam cohort. J Nutr. 2002;132:1335-40.
28. Sudi KM, Gallistl S, Tafeit E, Möller R, Borkenstein MH. The relationship between different subcutaneous adipose tissue layers, fat mass and leptin in obese children and adolescents. J Pediatr Endocrinol Metab. 2000;13:505-12.
29. Summerbell CD, Douthwaite W, Whittaker V, et al. The association between diet and physical activity and subsequent excess weight gain and obesity assessed at 5 years of age or older: a systematic review of the epidemiological evidence. Int J Obes (Lond). 2009;33(3 suppl):S1-92.
30. Thomas CD, Peters JC, Reed GW, Abumrad NN, Sun M, Hill JO. Nutrient balance and energy expenditure during ad libitum feeding of high-fat and high-carbohydrate diets in humans. Am J Clin Nutr. 1992;55:934-42.
31. Williams PT. Interactive effects of exercise, alcohol, and vegetarian diet on coronary heart disease risk factors in 9242 runners: the National Runners' Health Study. Am J Clin Nutr. 1997;66:1197-206.
32. Williams PT. Relationship of distance run per week to coronary heart disease risk factors in 8283 male runners. The National Runners' Health Study. Arch Intern Med. 1997;157:191-8.
33. Williams PT. Exercise and the population distribution of body weight. Int J Obes Relat Metab Disord. 2004;28:120-8.
34. Williams PT. Maintaining vigorous activity attenuates 7-yr weight gain in 8340 runners. Med Sci Sports Exerc. 2007;39(5):801-9.
35. Williams PT. Asymmetric weight gain and loss from increasing and decreasing exercise. Med Sci Sports Exerc. 2008;40(2):296-302.
36. Williams PT. Self-selection accounts for inverse association between weight and cardiorespiratory fitness. Obesity (Silver Spring). 2008;16:102-6.
37. Williams PT, Hoffman K, La I. Weight-related increases in hypertension, hypercholesterolemia, and diabetes risk in normal weight male and female runners. Arterioscler Thromb Vasc Biol. 2007;27:1811-9.
38. Williams PT, Hoffman KM. Optimal body weight for the prevention of coronary heart disease in normal-weight physically active men. Obesity (Silver Spring). 2009;17:1428-34.
39. Williams PT, Wood PD. The effects of changing exercise levels on weight and age-related weight gain. Int J Obes (Lond). 2006;30:543-51.
40. Zurlo F, Lillioja S, Esposito-Del Puente A, et al. Low ratio of fat to carbohydrate oxidation as predictor of weight gain: study of 24-h RQ. Am J Physiol. 1990;259(5 Pt 1):E650-7.


©2011The American College of Sports Medicine