According to the report of the World Health Organization (1985), total energy expenditure (TEE) in human subjects can be calculated as BMR×physical activity level (PAL). However, other reports have pointed out limitations in the suggested procedure related to the % body fat of the subjects. The purpose of the present study was to evaluate the World Health Organization (1985) procedure in thirty-four healthy women with BMI 18–39kg/m2. BMR and TEE were measured using indirect calorimetry (BMRmeas) and the doubly-labelled water method (TEEref) respectively. When assessed using the doubly-labelled water and skinfold-thickness methods, the women had 34 (sd 8) and 33 (sd 6) % body fat respectively. On the basis of guidelines provided by the World Health Organization (1985), 1·64 was selected to represent the average PAL of the women. Furthermore, PAL was also assessed by means of an accelerometer (PALacc), heart-rate recordings (PALHR) and a questionnaire (PALq). These estimates were: PALacc 1·71 (sd 0·17), PALHR 1·76 (sd 0·24), PALq 1·86 (sd 0·27). These values were lower than TEEref/BMRref, which was 1·98 (sd 0·21). BMR assessed using equations recommended by the World Health Organization (1985) (BMRpredicted) overestimated BMR by 594 (sd 431) kJ/24h. However, when TEE was calculated as BMRpredicted×PALacc, BMRpredicted×PALHR and BMRpredicted×PALq respectively, average results were in agreement with TEEref. Furthermore, TEE values based on BMRpredicted and PALacc, PALHR, PALq as well as on PAL=1·64, minus TEEref, were significantly correlated with body fatness. When the same PAL value (1·64) was used for all subjects, this correlation was particularly strong. Thus, the World Health Organization (1985) procedure may give TEE results that are biased with respect to the body fatness of subjects.