Background Previous reviews have identified 44 risk factors for poor early child development (ECD) in low‐ and middle‐income countries. Further understanding of their relative influence and pathways is needed to inform the design of interventions targeting ECD. Methods We conducted path analyses of factors associated with 18‐month language and motor development in four prospective cohorts of children who participated in trials conducted as part of the International Lipid‐Based Nutrient Supplements (iLiNS) Project in Ghana (n = 1,023), Malawi (n = 675 and 1,385), and Burkina Faso (n = 1,122). In two cohorts, women were enrolled during pregnancy. In two cohorts, infants were enrolled at 6 or 9 months. In multiple linear regression and structural equation models (SEM), we examined 22 out of 44 factors identified in previous reviews, plus 12 additional factors expected to be associated with ECD. Results Out of 42 indicators of the 34 factors examined, 6 were associated with 18‐month language and/or motor development in 3 or 4 cohorts: child linear and ponderal growth, variety of play materials, activities with caregivers, dietary diversity, and child hemoglobin/iron status. Factors that were not associated with child development were indicators of maternal Hb/iron status, maternal illness and inflammation during pregnancy, maternal perceived stress and depression, exclusive breastfeeding during 6 months postpartum, and child diarrhea, fever, malaria, and acute respiratory infections. Associations between socioeconomic status and language development were consistently mediated to a greater extent by caregiving practices than by maternal or child biomedical conditions, while this pattern for motor development was not consistent across cohorts. Conclusions Key elements of interventions to ensure quality ECD are likely to be promotion of caregiver activities with children, a variety of play materials, and a diverse diet, and prevention of faltering in linear and ponderal growth and improvement in child hemoglobin/iron status.