BACKGROUND: Few birth cohorts in South America evaluate the joint effect of minerals and toxic metals on neonatal health. In Madre de Dios, Peru, mercury exposure is prevalent owing to artisanal gold mining, yet its effect on neonatal health is unknown. OBJECTIVES: We aimed to determine whether toxic metals are associated with lower birth weight and shorter gestational age independently of antenatal care and other maternal well-being factors. METHODS: Data are from the COhorte de NAcimiento de MAdre de Dios (CONAMAD) birth cohort, which enrolled pregnant women in Madre de Dios prior to their third trimester and obtained maternal and cord blood samples at birth. We use structural equation models (SEMs) to construct latent variables for the maternal metals environment (ME) and the fetal environment (FE) using concentrations of calcium, iron, selenium, zinc, magnesium, mercury, lead, and arsenic measured in maternal and cord blood, respectively. We then assessed the relationship between the latent variables ME and FE, toxic metals, prenatal visits, hypertension, and their effect on gestational age and birth weight. RESULTS: Among 198 mothers successfully enrolled and followed at birth, 29% had blood mercury levels that exceeded the U.S. Centers for Disease Control and Prevention threshold of 5.8 μg/L and 2 mothers surpassed the former 5-μg/dL threshold for blood lead. The current threshold value is 3.5 μg/dL. Minerals and toxic metals loaded onto ME and FE latent variables. ME was associated with FE (β = 0.24; 95% CI: 0.05, 0.45). FE was associated with longer gestational age (β = 2.31; 95% CI: - 0.3, 4.51) and heavier birth weight. Mercury exposure was not directly associated with health outcomes. A 1% increase in maternal blood lead shortened gestational age by 0.05 d (β =  - 0.75; 95% CI: - 1.51, - 0.13), which at the 5-μg/dL threshold resulted in a loss of 3.6 gestational days and 76.5 g in birth weight for newborns. Prenatal care visits were associated with improved birth outcomes, with a doubling of visits from 6 to 12 associated with 5.5 more gestational days (95% CI: 1.6, 9.4) and 319 g of birth weight (95% CI: 287.6, 350.7). DISCUSSION: Maternal lead, even at low exposures, was associated with shorter gestation and lower birth weight. Studies that focus only on harmful exposures or nutrition may mischaracterize the dynamic maternal ME and FE. SEMs provide a framework to evaluate these complex relationships during pregnancy and reduce overcontrolling that can occur with linear regression.