The Pregnancy Obesity Nutrition and Child Health study is a longitudinal study of reproductive health. Here we analyzed body composition of normal-weight and obese Swedish women by three methods during each trimester of pregnancy. Cross-sectional and longitudinal fat mass estimates using quantitative magnetic resonance (QMR) and bioelectrical impedance analysis (BIA) (Tanita MC-180MA-III) were compared with fat mass determined by air displacement plethysmography (ADP) in pregnancy weeks 8-12, 24-26, and 35-37 in normal-weight women (n = 122, BMI = 22.1 ± 1.6 kg/m2) and obese women (n = 29, BMI = 34.6 ± 3.6 kg/m2). ADP results were calculated from pregnancy-adjusted fat-free mass densities. Mean fat mass by QMR and ADP were similar in obese women, although with wide limits of agreement. In normal-weight women, QMR overestimated mean fat mass in all trimesters, with systematic overestimation at low fat mass values in trimesters 1 and 3. In obese women, fat mass by BIA was grossly underestimated and imprecise in all trimesters, especially at higher values in trimester 2. In normal-weight women, fat mass by BIA was moderately lower than by ADP in trimester 1, similar in trimester 2, and moderately higher in trimester 3. QMR and ADP assessed fat mass changes similarly in obese women, whereas BIA overestimated fat mass changes in normal-weight women. Mean fat mass and fat mass changes by QMR and pregnancy-adjusted ADP were similar in pregnant obese women. Mean fat mass by QMR and fat mass changes by BIA were higher than corresponding values determined by pregnancy-adjusted ADP in normal-weight women.