Objectives Factors acting before children are born or reach school‐going age may explain why some do not complete primary education. Many relevant factors relate to maternal phenotype, but few studies have tested for independent associations of maternal factors relative to those characterizing the family in general. Methods Using data from a longitudinal study of 838 children in Dhanusha, Nepal, we used logistic regression models to test whether indices of maternal somatic and educational capital, or family economic capital, were independently associated with children having had ≤2 versus 3+ years of schooling at a mean age of 8.5 years. We also tested whether maternal age, children's early growth, and urban/rural location mediated such associations. Results Children had a higher risk of completing less schooling if their mothers were short, thin, anemic, and uneducated. Independently, lower family material assets and land acreage also increased children's odds of less schooling. There was an indication of gender differences, with the risk of poor educational attainment in girls associated with low maternal somatic and educational capital, whereas in boys the relevant factors were low maternal education and family land ownership. Conclusions Our analysis demonstrates that, independent of broader indices of family capital such as land or material assets, children's educational attainment is associated with factors embodied in maternal phenotype. Both somatic and educational maternal capital appeared important. A composite index of maternal capital could provide a new measurable proxy, prior to school entry, for identifying children at risk of completing fewer years of schooling. Am. J. Hum. Biol. 28:687–698, 2016. © 2016 Wiley Periodicals, Inc.