Abstract Background and objectives Women’s nutritional status is important for their health and reproductive fitness. In a population where early marriage is common, we investigated how women’s nutritional status is associated with their age at marriage (marking a geographical transfer between households), and at first pregnancy. Methodology We used data from a cluster-randomized control trial from lowland Nepal (n = 4071). Outcomes including body mass index (BMI) were measured in early pregnancy and trial endpoint, after delivery. We fitted mixed-effects linear and logistic regression models to estimate associations of age at marriage and age at pregnancy with outcomes, and with odds of chronic energy deficiency (CED, BMI <18.5 kg/m2), at both timepoints. Results BMI in early pregnancy averaged 20.9 kg/m2, with CED prevalence of 12.5%. In 750 women measured twice, BMI declined 1.2 (95% confidence interval 1.1, 1.3) kg/m2 between early pregnancy and endpoint, when CED prevalence was 35.5%. Early pregnancy was associated in dose-response manner with poorer nutritional status. Early marriage was independently associated with poorer nutritional status among those pregnant ≤15 years, but with better nutritional status among those pregnant ≥19 years. Conclusions and implications The primary determinant of nutritional status was age at pregnancy, but this association also varied by marriage age. Our results suggest that natal households may marry their daughters earlier if food insecure, but that their nutritional status can improve in the marital household if pregnancy is delayed. Marriage age therefore determines which household funds adolescent weight gain, with implications for Darwinian fitness of the members of both households.