Weight cycling of athletes and subsequent weight gain in middleage



To study the effects of repeated cycles of weight loss and regain as young adults on long-term weight development.


A follow-up study with questionnaires in 1985, 1995 and 2001.




A national cohort of 1838 male elite athletes who had represented Finland in major international sport competitions in 1920–1965, including 370 men engaged in sports in which weight-related performance classes are associated with weight cycling (boxers, weight lifters and wrestlers; further called as weight cyclers), and 834 matched control men with no athletic background.

Outcome measure:

Weight change since the age of 20 years, body mass index (BMI) and prevalence of obesity and overweight.


The weight cyclers gained 5.2 BMI units from age 20 years to their maximum mean weight, which was at age 58.7 years. Corresponding figures for the controls were 4.2 BMI units at 58.5 years and for other athletes 3.3 BMI units at age 62.5 years. The proportion of obese (BMI30 kg/m2) subjects was greatest among the weight cyclers both in 1985 and 1995. In 2001, the weight cyclers were more often obese than other athletes, but did not differ from the controls. The odds ratio for the weight cyclers to be obese compared to other athletes in 1985 was 3.18 (95% confidence intervals 2.09–4.83), and compared to the controls 2.0 (1.35–2.96). The enhanced weight gain of the weight cyclers was not accounted for by present health habits (smoking, alcohol use, use of high-fat milk or physical activity) or weight at age 20 years.


Repeated cycles of weight loss and regain appear to enhance subsequent weight gain and may predispose to obesity. Chronic dieting with weight cycling may be harmful for permanent weight control.


Chronic dieters and athletes engaged in sports with weight classes (e.g. wrestlers, weightlifters and boxers) periodically restrict their food intake to lose weight. The periods of food restriction and weight loss usually alternate with periods of unrestricted food intake and weight gain. The long-term effects of periodic dieting and weight regain on long-term weight development are still controversial.1, 2, 3

We therefore studied long-term weight changes in a unique population of former world-class elite athletes, including a group of men who had been active in sports with weight classes that have been linked to periodic weight reduction and regain practices. As our interest was in the development of overweight and obesity, we focused on changes in body weight in middle age. Later in life, body weight change is a function of both change in body fat and the senile decline of weight due to loss of fat-free mass, principally muscle mass.4 Therefore, we limited our analysis to include only the age range of 20– 65 years.

Subjects and methods

The study subjects were male athletes, who had represented Finland in the Olympic Games or in other major international sport competitions between 1920 and 1965 in track and field athletics, cross-country skiing, football, ice hockey, basketball, shooting, boxing, wrestling or weightlifting. Athletes were identified from sports yearbooks and registers and traced through the records of local parishes, and 97.7% of them were able to be traced. The control group consisted of conscripts from the same age cohort and area of residence as the athletes, and were selected from the service register of the Finnish Defence Forces; because of universal conscription, all Finnish men are assessed for eligibility to serve. The completely healthy conscript listed in the service register living the closest geographically to each athlete was chosen as his matched control. The controls were selected in 1978–1979, and no controls for athletes traced later were obtained.

The surviving former athletes and controls were sent in 1985, 1995 and 2001 an extensive questionnaire on personal characteristics, including height, current weight, weight at the time of their military service (at age 20 years), sociodemographic factors, psychological traits, smoking and the use of alcohol, physical activity, the discontinuation of their sporting career, dietary habits, symptoms and diseases.

Relative body weight was expressed as body mass index (BMI), which is weight in kilograms divided by height in meters squared. For each questionnaire, the current age was used and the current BMI was computed using the current weight and the consensus height as determined from all answers. For the BMI at age 20 years, we used the earliest available reported weight at age 20 years and the consensus height.

Habitual physical activity at leisure was assessed by a set of questions about the type, duration and frequency of various activities. The amount of energy expended in habitual physical activities was calculated and expressed in metabolic units, as described elsewhere.5 Alcohol consumption of all beverage types was computed as grams of alcohol per month and adjusted for body weight.5 A detailed smoking history was used to define smoking status as a categorical variable divided into never, former and current regular smokers. The group of occasional smokers was small (n=86) and similar to the former smokers in BMI and age, and thus was combined with them. Type of milk used, was used as an indicator variable of fat intake.6 Milk use was asked and used in analyses as a categorical variable with three categories: skimmed milk, medium-fat milk and high-fat milk.

Men who had been engaged in sports with mandatory weight classes (boxing, weightlifting and wrestling) in which weight reduction before a competition is common will be referred to as ‘weight cyclers’ in the text. The overall response rates were 83.6, 75.6 and 75.6% in 1985, 1995 and 2001, respectively. In 1985, the response rates for the controls, other athletes and the weight cyclers were respectively 91.5, 78.7 and 91.0%. In 1995, the corresponding figures were 81.6, 70.7 and 81.2%, and in 2001, 71.2, 82.3 and 78.5%. The total number of subjects aged less than 65 years in at least one questionnaire and with data on BMI at age 20 years was 2033, of which 273 (13.5%) were classified as weight cyclers, 1093 (53.7%) as other athletes and 667 (32.8%) as controls.

Validity assessment

Height and weight were measured for a sample of 87 athletes in 1992 participating in a laboratory study.7 Self-reported height and weight agreed closely to measured height and weight (for height, r=0.96 (1985 questionnaire) and r=0.98 (1995), and for weight, r=0.90 (1985) and r=0.94 (1995)).

To analyze a possible recall bias in reporting body weight at 20 years, we examined the correlations between reported weights on different questionnaires. The correlations were greater than 0.91 between all time points.

Statistical methods

The effect of weight cycling status on weight gain and BMI was analyzed with linear regression using weight gain and BMI as outcome variables in separate models for each questionnaire. The prevalence of overweight only (BMI 25–29.9 kg/m2) and obesity (BMI30 kg/m2) were age-adjusted by direct standardization to the age-distribution of the entire sample. The risk of overweight only and obesity were analyzed using logistic regression models. All the models were adjusted for age at the time of the questionnaire and BMI at age 20 years and limited to subjects aged less than 65 years. To examine if the differences in weight gain were due to differences in life habits, we adjusted the previous models for smoking, alcohol use, type of milk used and physical activity. To further examine the effect of smoking on weight gain, we used stratified models according to smoking status and weight cycling status separately, and then tested for smoking by weight cycling status interactions. As the use of alcohol and physical activity variables were highly skewed, we used a logarithmic (log+1) transformation of them in the regression analysis. For the validity assessment, we calculated intraclass correlations between measured and reported heights and weights.

Because age was an important explanatory factor for weight gain, we examined the age dependency of BMI using information on BMI from all questionnaires in the regression models for repeated measures, again limited to age less than 65 years. In these models, the interaction between age and weight cycling status was tested comparing models with and without interaction terms with a likelihood ratio test. Based on the regression equations in these models, we estimated change in BMI from age 20 years to the age of the maximum average BMI for each study group (weight cyclers, other athletes and controls). All statistical analyses were carried out with the Stata statistical software version 8.2.8


In 1985, the weight cyclers were older (P<0. 05) than the other men, and in 1995, other athletes were younger (P<0.005) than the other men. In 2001, there was no longer an age difference between groups. The number of observations also varies between time points because the questionnaire was sent at every round to all surviving subjects included in the cohort independent of their previous response status (Table 1).

Table 1 The number of subjects, mean age, age-adjusted BMI and weight change since age 20 years by group in different study years of subjects aged 65 years and under at the time of the questionnaire

At age 20 years, the BMI did not significantly differ between groups. In 1985 and 1995, the age-adjusted mean BMI of the weight cyclers significantly exceeded that of the control men, which was higher than the mean BMI of the other former sportsmen. In 2001, the difference in age-adjusted mean BMI disappeared between the weight cyclers and controls, but remained statistically significant between the weight cyclers and other athletes (Table 1).

In 1985 and 1995, the weight cyclers had the greatest mean weight gain (in 1985, 15.1 kg, 95% confidence intervals (CI) 13.3–17.0) compared to other athletes (9.6 kg, 8.5–10.6) or the controls (11.8 kg, 10.5–13.1) compared to weight at age 20 years. In 2001, the weight cyclers and controls both had a mean weight gain of over 14 kg, whereas the result for other athletes was only 10.4 kg (Table 1). Stratification for smoking status did not change the results for the weight cyclers, but revealed that current smoking controls did not gain weight, whereas never and former smoker controls did (P<0.01 between former and current smokers, P=0.14 between current and never smokers) (data not shown).

The weight cyclers were significantly more likely to be obese than other athletes at all time points (odds ratio (OR) 2.35–5.05). Compared to the controls, weight cyclers were twice as likely to be obese in 1985 and 1995, but no difference was found in the relative risk of obesity in 2001 (Table 2). The evolution over time of the weight cyclers' relative risk was not due to the composition of the study group; the same result was seen when only those who had responded to all three analyses were considered. These differences in obesity were also observed when the data were stratified by smoking status (data not shown).

Table 2 Odds ratio (95% CI) and age-adjusted prevalence for overweight (BMI 25–29.9 kg/m2) and obesity (BMI30kg/m2), of weight cyclers vs other athletes and controls aged 65 and under at the time of the questionnaire

Some differences were seen in life habits between the groups. Between all the groups, the controls were most likely to be current or former smokers (P<0.001), and when the two athlete groups are compared, the weight cyclers were more likely to be current smokers than other athletes (P<0.05). The alcohol use of the weight cyclers did not differ from that of the controls (P>0.05). Other athletes were more physically active than the weight cyclers, who were more active than the controls (P<0.05 between groups). The weight cyclers drank high-fat milk more often than other athletes but less often compared to the controls (P<0.05 between groups) (Table 3). Adjustment for life habits (smoking, alcohol use, type of milk used and physical activity) did not substantially change the likelihood of obesity (e.g. OR for obesity at 1985 for the weight cyclers vs other athletes before and after adjustment: 3.18 (95% CI 2.09–4.83) vs 2.74 (1.47–5.10) or the weight cyclers compared to controls 2.0 (1.35–2.96) adjusted 1.52 (0.86–2.69)) or weight gain (detailed data not shown).

Table 3 Age and health habits by weight cycling category in 1985 distributed by categories (percentages) or means (only subjects aged under 65 years)

Using information from all time points, the regression of BMI on age was a quadratic function such that both age and age2 predicted BMI (Figure 1). The mean BMI as a function of age rises for all groups, but rises more steeply for the weight cyclers, and then begins to slope downwards at the oldest ages (Figure 1). This interaction of weight cycling status and age was statistically significant (P<0.001). Based on the regression equations, the estimated increase in weight from age 20 years to its maximum value was 5.2 BMI units (at age 58.7 years) for the weight cyclers and 4.2 BMI units (at age 58.5 years) for the controls. The increase for other athletes was smaller (3.3 BMI units) with a maximum at a later age (62.5 years).

Figure 1

Mean body mass index (BMI) and 95% confidence intervals (CI) as a function of age.


Our finding of a significantly greater weight gain for the weight cyclers (boxers, wrestlers and weightlifters) at a younger age than for other sportsmen or for the control men could not be attributed to differences in current health habits, even though the lifestyle of the former athletes was much healthier5 than that of the controls. It is therefore possible that weight loss practices during the active sporting career of the weight cyclers could have modified their subsequent weight development, increasing the propensity to weight gain after repeated cycles of weight loss and regain.

The health effects of weight cycling have been subject to active research and sometimes passionate discussion both in the scientific literature and the public media for years. A recent large population-based study (a nationally representative cohort follow-up in the US, NHANES I; n=8479) reported a higher all cause and cardiovascular disease (CVD) mortality of weight cyclers compared to non-obese stable weight subjects.9 Similar results were reported by Hamm et al.10 for a 25-year follow-up of a large male cohort and Lissner et al.11 in a large population-based study. Studies on smaller samples and intervention studies have contradictory results about the effect of weight cycling on CVD mortality and morbidity.12, 13, 14

Weight cycling has also been attributed to greater weight gain and binge eating,15, 16, 17, 18, 19 psychological stress and lower general well-being20 and unhealthier body fat distribution.15, 21 However, Prentice et al.22 did not find any detrimental effects of weight cycling on body composition in a large African population-based study nor in an experimental prospective study of moderately obese women. In a cross-sectional setting of active young wrestlers aged 16 years, Steen et al.23 did not find differences in body size, fat or lean body mass between weight cyclers and non-cyclers.

We could not directly assess possible weight cycling behavior during the sport career in our sample, but weight cycling and unhealthy weight control methods (‘weight cutting’) have been widely reported in young male athletes participating in wrestling24, 25, 26, 27, 28 and other sports with weight classes.29 Over 80% of high school25 and college26 wrestlers have been reported to ‘cut weight’, and 26.6% to cut more than 10 times per season. The effect of weight cutting is usually a loss of several kilograms, which is quickly regained.26, 27 We also did not have measurements of the body composition of our subjects and therefore could not evaluate changes in body composition over time as their weight changed. The interpretation of weight gain in the athletes over time is more complex than for non-athletes, who typically have much less muscle mass. To be able to distinguish the obese from the only overweight muscular subjects, we studied the BMI distribution in the categories of overweight only (BMI 25–29.9 kg/m2) and obesity (BMI30 kg/m2). Our finding of much higher obesity prevalence in weight cyclers than other groups indicates major weight gain that cannot be attributed only to the greater muscle mass of athletes in these sports.

Limiting the data to subjects aged less than 65 years separately at each survey causes the exclusion of older cohorts from the original sample. Likewise, differential mortality due to obesity may have affected results, as the sportsmen and controls age in this cohort study; this may account for the disappearance in excess risk at the last survey (in 2001) compared to earlier time points. Taking into account the nearly 20-year time span of this study, there might be some period-specific effects on weight gain in our sample, given the general trend for gain in the entire Finnish population.30 The rate of weight gain in adulthood has increased in men, but not in women, from 1972 to 1997.30 Hence, the effects of the weight cycling of present-day athletes and other young adult men on future weight gain cannot be predicted with certainty. Moreover, our sample is restricted to men, and we cannot generalize to women.

Repeated cycles of weight loss and regain have in some but not in all animal studies been shown to lower resting metabolic rate31 or feed efficiency.2 Similarly, a lower resting metabolic rate has been observed in adolescent wrestlers engaged in weight cycling than in non-cycling wrestlers in some23 but not in other32 studies. A reduced metabolic rate has also been observed in women engaged in cycles of binge eating and strict dieting,33 in restrained eaters34 and in post-obese individuals,35 although the findings are not unequivocal.31

A low metabolic rate reduces energy expenditure and together with binge eating can be expected to increase the propensity for weight gain. An increased risk of weight gain has been demonstrated in subjects with a low metabolic rate among Pima Indians, an indigenous population with a very high prevalence of obesity.31, 36 Weight cycling is associated with increased risk of binge eating,15, 18, 37 but we did not have any information about the eating patterns of the men in this study. Further, binge eating is associated to greater energy intake in some38, 39 but not in all40 studies. In our study, the weight cyclers appeared to have an enhanced preference for high-fat milk, which is in keeping with the observation that weight cycling may increase experimental animals' preference for fatty foods.2, 31 The greater than expected weight gain of the weight cycling athletes could perhaps be explained by a lower than expected basal energy expenditure, even if these men had maintained a physically active lifestyle compared to the control men. Another possible or supplementary explanation might be increased energy intake due to increased meal size and energy density associated with bingeing as a consequence of repeated weight cycle periods. Studies on eating behavior of athletes would be needed to further elucidate this problem.

The weight cycling behavior of the former athletes engaged in power sports at a young age resembles that of young dieters who lose weight temporarily and soon regain it. The present observations concerning the enhanced weight gain of these athletes raise the concern that the repeated cycles of weight loss and regain caused by dieting at a young age could similarly affect weight in the long term. For many young persons with perceived rather than real excess weight, a temporary success in the relentless pursuit for thinness may predict an ever increasing problem with weight maintenance as well as an increased risk of eating disorders. The discouragement of unnecessary weight control measures may be one of the means to prevent obesity and eating disorders.


  1. 1

    Kroke A, Liese AD, Schulz M, Bergmann MM, Klipstein-Grobusch K, Hoffmann K et al. Recent weight changes and weight cycling as predictors of subsequent two year weight change in a middle-aged cohort. Int J Obes Relat Metab Disord 2002; 26: 403–409.

  2. 2

    Reed DR, Contreras RJ, Maggio C, Greenwood MR, Rodin J . Weight cycling in female rats increases dietary fat selection and adiposity. Physiol Behav 1988; 42: 389–395.

  3. 3

    Wannamethee SG, Shaper AG, Walker M . Weight change, weight fluctuation, and mortality. Arch Intern Med 2002; 162: 2575–2580.

  4. 4

    Ryan AS, Elahi D . Body:composition, weight, height and build. In: Birren JE (ed). Encyclopedia of Gerontology. Academic Press: San Diego,CA, 1996. pp. 193–201.

  5. 5

    Fogelholm M, Kaprio J, Sarna S . Healthy lifestyles of former Finnish world class athletes. Med Sci Sports Exerc 1994; 26: 224–229.

  6. 6

    Roos E, Ovaskainen M-L, Pietinen P . Validity and comparison of three saturated fat indices. Scand J Nutr 1995; 39: 55–59.

  7. 7

    Kettunen JA, Kujala UM, Raty H, Videman T, Sarna S, Impivaara O et al. Factors associated with hip joint rotation in former elite athletes. Br J Sports Med 2000; 34: 44–48.

  8. 8

    Statacorp. Stata Statistical Software: Release 8.2. Stata Corporation: College Station, TX, 2003.

  9. 9

    Diaz VA, Mainous AG, Everett CJ . The association between weight fluctuation and mortality: results from a population-based cohort study. J Commun Health 2005; 30: 153–165.

  10. 10

    Hamm P, Shekelle RB, Stamler J . Large fluctuations in body weight during young adulthood and twenty-five-year risk of coronary death in men. Am J Epidemiol 1989; 129: 312–318.

  11. 11

    Lissner L, Odell PM, D'Agostino RB, Stokes III J, Kreger BE, Belanger AJ et al. Variability of body weight and health outcomes in the Framingham population. N Engl J Med 1991; 324: 1839–1844.

  12. 12

    Wing RR, Jeffery RW, Hellerstedt WL . A prospective-study of effects of weight cycling on cardiovascular risk-factors. Arch Int Med 1995; 155: 1416–1422.

  13. 13

    Petersmarck KA, Teitelbaum HS, Bond JT, Bianchi L, Hoerr SM, Sowers MF . The effect of weight cycling on blood lipids and blood pressure in the Multiple Risk Factor Intervention Trial Special Intervention Group. Int J Obes Relat Metab Disord 1999; 23: 1246–1255.

  14. 14

    Kajioka T, Tsuzuku S, Shimokata H, Sato Y . Effects of intentional weight cycling on non-obese young women. Metabolism 2002; 51: 149–154.

  15. 15

    Field AE, Manson JE, Taylor B, Willett WC, Colditz GA . Association of weight change, weight control practices, and weight cycling among women in the Nurses' Health Study II. Int J Eat Disord 2004; 35: 394.

  16. 16

    Marchesini G, Cuzzolaro M, Mannucci E, Dalle GR, Gennaro M, Tomasi F et al. Weight cycling in treatment-seeking obese persons: data from the QUOVADIS study. Int J Obes Relat Metab Disord 2004; 28: 1456–1462.

  17. 17

    Womble LG, Williamson DA, Martin CK, Zucker NL, Thaw JM, Netemeyer R et al. Psychosocial variables associated with binge eating in obese males and females. Int J Eat Disord 2001; 30: 217–221.

  18. 18

    Venditti EM, Wing RR, Jakicic JM, Butler BA, Marcus MD . Weight cycling, psychological health, and binge eating in obese women. J Consult Clin Psychol 1996; 64: 400–405.

  19. 19

    Field AE, Manson JE, Laird N, Williamson DF, Willett WC, Colditz GA . Weight cycling and the risk of developing type 2 diabetes among adult women in the United States. Obes Res 2004; 12: 267–274.

  20. 20

    Foreyt JP, Brunner RL, Goodrick GK, Cutter G, Brownell KD, Stjeor ST . Psychological correlates of weight fluctuation. Int J Eat Disord 1995; 17: 263–275.

  21. 21

    Wallner SJ, Luschnigg N, Schnedl WJ, Lahousen T, Sudi K, Crailsheim K et al. Body fat distribution of overweight females with a history of weight cycling. Int J Obes Relat Metab Disord 2004; 28: 1143–1148.

  22. 22

    Prentice AM, Jebb SA, Goldberg GR, Coward WA, Murgatroyd PR, Poppitt SD et al. Effects of weight cycling on body composition. Am J Clin Nutr 1992; 56 (1 Suppl): 209S–216S.

  23. 23

    Steen SN, Oppliger RA, Brownell KD . Metabolic effects of repeated weight loss and regain in adolescent wrestlers. JAMA 1988; 260: 47–50.

  24. 24

    Lakin JA, Steen SN, Oppliger RA . Eating behaviors, weight loss methods, and nutrition practices among high school wrestlers. J Community Health Nurs 1990; 7: 223–234.

  25. 25

    Tipton CM, Tcheng TK . Iowa wrestling study. Weight loss in high school students. JAMA 1970; 214: 1269–1274.

  26. 26

    Oppliger RA, Steen SA, Scott JR . Weight loss practices of college wrestlers. Int J Sport Nutr Exerc Metab 2003; 1: 29–46.

  27. 27

    Alderman BL, Landers DM, Carlson J, Scott JR . Factors related to rapid weight loss practices among international-style wrestlers. Med Sci Sports Exerc 2004; 36: 249–252.

  28. 28

    Kiningham RB, Gorenflo DW . Weight loss methods of high school wrestlers. Med Sci Sports Exerc 2001; 33: 810–813.

  29. 29

    Fogelholm M, Hiilloskorpi H . Weight and diet concerns in Finnish female and male athletes. Med Sci Sports Exerc 1999; 31: 229–235.

  30. 30

    Lahti-Koski M, Jousilahti P, Pietinen P . Secular trends in body mass index by birth cohort in eastern Finland from 1972 to 1997. Int J Obes Relat Metab Disord 2001; 25: 727–734.

  31. 31

    National Task Force on the Prevention and Treatment of Obesity. Weight cycling. JAMA 1994; 272: 1196–1202.

  32. 32

    McCargar LJ, Crawford SM . Metabolic and anthropometric changes with weight cycling in wrestlers. Med Sci Sports Exerc 1992; 24: 1270–1275.

  33. 33

    Devlin MJ, Walsh BT, Kral JG, Heymsfield SB, Pi-Sunyer FX, Dantzic S . Metabolic abnormalities in bulimia nervosa. Arch Gen Psychiatry 1990; 47: 144–148.

  34. 34

    Poehlman ET, Viers HF, Detzer M . Influence of physical activity and dietary restraint on resting energy expenditure in young nonobese females. Can J Physiol Pharmacol 1991; 69: 320–326.

  35. 35

    Leibel RL, Hirsch J . Diminished energy requirements in reduced-obese patients. Metabolism 1984; 33: 164–170.

  36. 36

    Hanson RL, Jacobsson LTH, McCance DR, Narayan KMV, Pettitt DJ, Bennett PH et al. Weight fluctuation, mortality and vascular disease in Pima Indians. Int J Obes Relat Metab Disord 1996; 20: 463–471.

  37. 37

    Fairburn CG, Wilson GT (eds). Binge Eating. Nature, Assessment and Treatment. The Guilford Press: New York, 1993.

  38. 38

    Goldfein JA, Walsh BT, LaChaussee JL, Kissileff HR, Devlin MJ . Eating behavior in binge eating disorder. Int J Eat Disord 1993; 14: 427–431.

  39. 39

    Yanovski SZ, Sebring NG . Recorded food intake of obese women with binge eating disorder before and after weight loss. Int J Eat Disord 1994; 15: 135–150.

  40. 40

    Gosnell BA, Mitchell JE, Lancaster KL, Burgard MA, Wonderlich SA, Crosby RD . Food presentation and energy intake in a feeding laboratory study of subjects with binge eating disorder. Int J Eat Disord 2001; 30: 441–446.

Download references


SES has received financial support from Yrjö Jahnsson foundation (5074), and has been supported by the doctoral programs in Public Health graduate school. This study is part of the Centre of Excellence in Complex Disease Genetics funded by the Academy of Finland, and has been supported by the GenomEUtwin project (European Union Contract No. QLG2-CT-2002-01254).

Author information

Correspondence to S E Saarni.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Saarni, S., Rissanen, A., Sarna, S. et al. Weight cycling of athletes and subsequent weight gain in middleage. Int J Obes 30, 1639–1644 (2006). https://doi.org/10.1038/sj.ijo.0803325

Download citation


  • body weight changes
  • weight gain
  • athletics
  • overweight

Further reading