Objective. To examine the association between maternal obesity and infant mortality, while including information about mode of delivery and interpregnancy weight change. Design. Register-based cohort study. Setting and population. A total of 1 199 183 singletons, including 3481 infant deaths, from the Swedish Birth Register 1992–2006. Methods. Maternal body mass index (BMI) was obtained from self-reports in early pregnancy and categorized as underweight (<18.5 kg/m2), normal-weight (18.5–24.9 kg/m2), overweight (25–29.9 kg/m2), obese (30–34.9 kg/m2) and extremely obese (≥35 kg/m2). Cox regression was used to estimate hazard ratios (95% confidence intervals). Infants of normal-weight women were the referent. Main outcome measures. Neonatal and infant mortality. Results. Infant mortality increased with increasing maternal fatness [adjusted hazard ratios 1.2 (1.1–1.3), 1.4 (1.2–1.6) and 2.1 (1.8–2.5) for overweight, obesity and extreme obesity, respectively]. When accounting for mode of delivery, neonatal mortality was increased in infants of obese and extremely obese mothers after spontaneous births [adjusted hazard ratios 1.8 (1.4–2.4) and 2.6 (1.8–4.0), respectively, after term births, and 1.4 (1.1–1.9) and 2.2 (1.5–3.3), respectively, after preterm births]. No excess risk was present for infants of obese mothers after induced term and preterm births (p-values for interaction <0.05). For post-neonatal mortality, no interaction between mode of delivery and maternal obesity was observed. In women with two subsequent pregnancies, high interpregnancy weight change >1 BMI unit (1 kg/m2) seemed to involve a modest increase in neonatal mortality in the second infant, but only after spontaneous births [adjusted odds ratio 1.3 (0.9–1.7)]. Conclusions. Maternal obesity, especially at levels that may involve cardiometabolic morbidity, was associated with increased mortality in the offspring.