Introduction  Similarities in the symptomatic expressions of excess adiposity and hypercortisolaemic conditions suggest that elevated glucocorticoid exposure may influence the pathogenesis of obesity. Circulating cortisol levels are not typically elevated in obese subjects, but data from large prospective samples are rare. We undertook an analysis to determine both cross‐sectional and longitudinal associations between body composition and serum cortisol concentrations in a randomly chosen group of 999 community‐dwelling men, aged 40–79 years.

Methods  Data were obtained from the two follow‐up waves of the Massachusetts Male Ageing Study (T2: 1995–97; T3: 2002–04). Partial correlation and multivariate regression analyses were used to estimate cross‐sectional (T2) and longitudinal associations between serum cortisol concentrations and a range of measures of subjects’ body composition, including weight, body mass index (BMI), waist circumference (WC), waist‐to‐hip girth ratio (WHR), and percentage body fat (measured by bioelectrical impedance at T3); similar analyses were conducted to assess the association between change (T2 to T3) in serum cortisol and simultaneous change in body composition parameters.

Results  We observed weak negative associations between cortisol concentrations and all body composition parameters, with the exception of percentage body fat. Longitudinal results demonstrated similar relationships but associations were of lesser magnitude. T2 cortisol concentrations were not associated with change in body composition over time, whereas T2 body size was positively associated with longitudinal changes in cortisol concentrations, providing limited evidence that weight change drives changes in cortisol concentrations, rather than vice versa. Results were unchanged when age and other covariate effects were controlled.

Conclusions  Circulating cortisol concentrations are somewhat lower in obese than in nonobese community‐dwelling men. There is some evidence that excess adiposity presages increases in cortisol concentrations, rather than the reverse. However, this observation should be greeted with caution, as age‐related weight loss – and not gain – was associated with simultaneous increases in serum cortisol concentrations.