Previous epidemiological studies have suggested an association between low levels of physical activity, fitness and the metabolic cardiovascular syndrome. However, many studies have used subjective non-quantitative questionnaire-based methods for assessing physical activity which do not distinguish between the different dimensions of this complex exposure, and in which measurement error in the exposure has not been estimated. These deficiencies in the measurement of this exposure complicate the interpretation of the results of epidemiological studies, and consequently make it difficult to design appropriate interventions and to estimate the expected benefit which would result from intervention. In particular, it is unclear whether public health advice should be to increase total energy expenditure, or to attempt to raise fitness by recommending periods of vigorous activity. To separate the effects of fitness and total energy expenditure in the aetiology of the metabolic cardiovascular syndrome, we measured the physical activity level (PAL), defined as total energy expenditure: BMR, and fitness (maximum O2 consumption (Vo2max per kg), measured in a sub-maximal test) in a cross-sectional population-based study of 162 adults aged 30–40 years. Heart-rate monitoring with individual calibration was used to measure total energy expenditure using the HRFlex method (Ceesay et al. 1989) which has been validated previously against doubly-labelled water and whole-body calorimetry. The relationship between a single measure of PAL, Vo2max per kg and the usual or habitual level for each exposure was measured in a sub-study of twenty-two subjects who undertook four repeated measures over the course of 1 year. This study design allows the reliability coefficient to be computed, which is used to adjust the observed associations for measurement error in the exposure. Twelve men (16.4%) and sixteen women (18.0%) were defined as having one or more features of the metabolic cardiovascular syndrome. The univariate odds ratio for each increasing quartile for PAL was 0.64 (95 % CI 0.43–0.94) and was 0.49 (95 % CI 0.32–0.74) for Vo2max per kg, suggesting that the association with the metabolic cardiovascular syndrome was stronger for fitness than for PAL. However, after adjustment for obesity and sex, and correction for exposure measurement error, the odds ratio per quartile for PAL was 0.32 (95 % CI 0.13–0.83) and 0.44 (95 % CI 0.24–0.78) for Vo2max per kg. Thus, although univariate analysis would suggest that fitness has a stronger association with the metabolic cardiovascular syndrome than PAL, this conclusion is reversed once confounding and the differences in measurement error are considered. We conclude from the present study that the metabolic cardiovascular syndrome is strongly associated with reduced habitual energy expenditure. The method employed to assess the exposure in the present study demonstrates the utility of assessing a known dimension of physical activity using a physiologically-based and objective measure with repeated estimation to adjust for measurement error. Such quantitative epidemiological data provide the basis for planning and evaluating the expected benefit of population-level interventions.