Peak oxygen uptake (VO2peak) is commonly indexed by total body weight (TBW) to determine cardiopulmonary fitness (CPF). This approach may lead to misinterpretation, particularly in obese subjects. We investigated the normalization of VO2peak by different body composition markers. We analyzed combined data of 3848 subjects (1914 women; 49.7%), aged 20‐90, from two independent cohorts of the population‐based Study of Health in Pomerania (SHIP‐2 and SHIP‐TREND). VO2peak was assessed by cardiopulmonary exercise testing. Body cell mass (BCM), fat‐free mass (FFM), and fat mass (FM) were determined by bioelectrical impedance analysis. The suitability of the different markers as a normalization variable was evaluated by taking into account correlation coefficients (r) and intercept (α‐coefficient) values from linear regression models. A combination of high r and low α values was considered as preferable for normalization purposes. BCM was the best normalization variable for VO2peak (r = .72; P ≤ .001; α‐coefficient = 63.3 mL/min; 95% confidence interval [CI]: 3.48‐123) followed by FFM (r = .63; P ≤ .001; α‐coefficient = 19.6 mL/min; 95% CI: −57.9‐97.0). On the other hand, a much weaker correlation and a markedly higher intercept were found for TBW (r = .42; P ≤ .001; α‐coefficient = 579 mL/min; 95% CI: 483 to 675). Likewise, FM was also identified as a poor normalization variable (r = .10; P ≤ .001; α‐coefficient = 2133; 95% CI: 2074‐2191). Sex‐stratified analyses confirmed the above order for the different normalization variables. Our results suggest that BCM, followed by FFM, might be the most appropriate marker for the normalization of VO2peak when comparing CPF between subjects with different body shape.