BackgroundLactation has been widely associated with optimal neurocognitive development, but the underlying mechanism remains unknown. Human milk oligosaccharides (HMOs) are complex sugars that support brain development, but previous studies examining their associations with cognition have yielded inconsistent findings. ObjectivesThis study aimed to provide a broader understanding of how HMOs jointly influence cognition. MethodsWe used data from an ongoing longitudinal cohort of Latino mother–infant dyads. Human milk samples from 1 mo (n = 157) and 6 mo (n = 107) postpartum were assessed for the 19 most abundant HMOs. Cognitive performance was assessed at 2 y using the Bayley Scale of Infant and Toddler Development. A partial least squares model identified HMO combinations predictive of cognitive scores. ResultsAt 1 mo, the combination of higher concentrations of lacto-N-neotetraose (LNnT), lacto-N-tetraose (LNT), lacto-N-fucopentaose (LNFP)-III, 6′-sialyllactose, and 2′-fucosyllactose (FL) with lower concentrations of sialyllacto-N-tetraose (LST) b, LNFP-II, fucodisialyllacto-N-hexaose, and 3-FL significantly predicted higher cognitive scores (β: 0.61; 95% confidence interval [CI]: 0.30, 0.92), explaining an additional 8% of the variance over a model with only nuisance covariates (11%). Additional analyses revealed that the combination of higher LNFP-III and lower LSTb alone explained 5% more of the variation in cognitive scores (β: 0.66; 95% CI: 0.24, 1.09). At 6 mo (n = 107), higher LNnT, LNT, and LNFP-III and lower 3FL and LSTb concentrations explained an extra 6% of the variance in cognitive scores (β: 0.43; 95% CI: 0.12, 0.75). ConclusionsThis study highlights specific HMO combinations in early life influencing cognitive performance at 2 y.