Genome wide association studies have identified ~100 loci associated with body mass index (BMI). Persons with low birth weight have an increased risk of metabolic disorders. We postulate that normal mechanisms of body weight regulation are disrupted in subjects with low birth weight. The present analyses included 2215 African American women from the Black Women's Health Study, and were based on genotype data on 20 BMI-associated loci and self-reported data on birth weight, weight at age 18 and adult weight. We used general linear models to assess the association of individual single-nucleotide polymorphisms (SNPs) with BMI at age 18 and later in adulthood within strata of birth weight (above and below the median, 3200 g). Three SNPs (rs1320330 near TMEM18, rs261967 near PCSK1 and rs17817964 in FTO), and a genetic score combining these three variants, showed significant interactions with birth weight in relation to BMI. Among women with birth weight <3200 g, there was an inverse association between genetic score and BMI; beta-coefficient=-0.045 (95% confidence intervals (CI) -0.104, 0.013) for BMI at age 18, and -0.055 (95% CI -0.112, 0.002) for adult BMI. Among women with birth weight ⩾3200 g, genetic score was positively associated with BMI: beta-coefficient=0.110 (95% CI 0.051, 0.169) for BMI at age 18 (P for interaction=0.0002), and 0.112 (95% CI 0.054, 0.170) for adult BMI (P for interaction<0.0001). Because TMEM18, PCSK1 and FTO are highly expressed in the central nervous system (CNS), our results suggest that low-birth weight may disrupt mechanisms of CNS body weight regulation.