The aim of the present study was to compare variables of metabolism, physical activity and fitness to body composition in normal and overweight children in a cross-sectional study design. Body composition was assessed by anthropometric measurements and bioelectrical impedance analysis in forty-eight prepubertal children (age 5–11 years, thirteen normal-weight, thirty-five overweight). Total energy expenditure (EE) was measured by combination of indirect calorimetry (for measurement of resting EE) and individually calibrated 24 h heart-rate (HR) monitoring. Activity-related EE and physical activity level (PAL) were calculated. Time spent with min-by-min HR>FLEX HR was also used as a marker of moderate habitual and vigorous activities. Aerobic fitness (O2 pulse (O2 consumption:HR at submaximal steady-state heart rate), submaximal O2 consumption (VO2submaximal), RER at a HR of 170 beats per min) was determined by bicycle ergometry. Muscle strength of the legs (maximal isometric strength of musculus quadriceps and of musculus ischiocruralis (Fa max and Fb max respectively)) was measured by computer tensiometry. When compared with normal children, overweight children had higher skinfold thicknesses (sum of skinfold thicknesses at four sites +160%), fat mass (+142%), waist (+24%) and hip circumferences (+14%), resting EE (+13%) and RER (+5%). No significant group differences were found for fat-free mass, muscle mass, total EE, activity-related EE, PAL, HR>FLEX HR, VO2submaximal, O2 pulse, Fa max and Fb max as well as the fat-free mass- or muscle mass-adjusted values for resting EE, aerobic fitness and muscle strength. When compared with normal children, overweight children had a lower measured v. estimated resting EE (Δ resting EE) and spent more time watching television. There were positive relationships between fat-free mass(x) and resting EE(x), total EE(y), aerobic fitness(y) and muscle strength(y), but only Δ resting EE(x) and HR>FLEX HR(x) correlated with fat mass(y). In a stepwise multivariate regression analysis resting EE adjusted for fat-free mass and Δ resting EE were significant determinants of % fat mass and explained 29·7% of its variance. Thus, in the present cross-sectional study, resting EE was the most important determinant of fat mass.