The Trivers–Willard hypothesis predicts the unequal parental investment between daughters and sons, depending on maternal condition and offspring reproductive potential. Specifically, in polygynous populations where males have higher reproductive variance than females, it predicts that mothers in good condition will invest more in sons, whereas mothers in poor condition will invest more in daughters. Previous studies testing this hypothesis focused on behavioral investment, whereas few examined biological investment. This study investigates the Trivers–Willard hypothesis on both behavioral and biological parental investment by examining breastfeeding frequencies and breast milk fat concentrations. Data from exclusively breastfeeding mothers in Northern Kenya were used to test hypotheses: Economically sufficient mothers will breastfeed sons more frequently than daughters, whereas poor mothers will breastfeed daughters more frequently than sons, and economically sufficient mothers will produce breast milk with higher fat concentration for sons than daughters, whereas poor mothers will produce breast milk with higher fat concentration for daughters than sons. Linear regression models were applied, using breastfeeding frequency or log-transformed milk fat as the dependent variable, and offspring's sex (son = 1/daughter = 0), socioeconomic status (higher = 1/lower = 0), and the sex-wealth interaction as the predictors, controlling for covariates. Our results only supported the milk fat hypothesis: infant's sex and socioeconomic status interacted (P = 0.014, n = 72) in their relation with milk fat concentration. The model estimated that economically sufficient mothers produced richer milk for sons than daughters (2.8 vs. 0.6 gm/dl) while poor mothers produced richer milk for daughters than sons (2.6 vs. 2.3 gm/dl). Further research on milk constituents in relation to offspring's sex is warranted. Am J Phys Anthropol , 2012. © 2012 Wiley Periodicals, Inc.