Background Inflammation has long-term health impacts across the life course. Breastfeeding substantially reduces inflammation risk, but key pathways, including the extent that this is due to protection against early life infection, are poorly understood. We aimed to investigate the relationships between breastfeeding, inflammation, and infection burden, and to determine the extent to which metabolomic and lipidomic profiles associated with breastfeeding mediate these health outcomes. Methods We utilised data from the Barwon Infant Study (BIS), a longitudinal birth cohort in Victoria, Australia. Infants (n = 889) with available breastfeeding (categorised as yes/no) clinical, metabolomic, and Lipidomic data at 6 and/or 12 months were included (n = 793 at 6 months, n = 734 at 12 months). Inflammation, measured via glycoprotein acetyls (GlycA), at 6 and 12 months and infection burden, including parent-reported and medically attended infections assessed through standardised 3-monthly questionnaires were used as outcomes. Results Any breastfeeding, regardless of supplementary feeding, was associated with lower inflammation, fewer infections, and significant, potentially beneficial changes in metabolomic and lipidomic markers, particularly plasmalogens. There was evidence of bidirectional mediation: metabolomic biomarkers and lipids mediated breastfeeding’s effects on inflammation, while inflammation partly mediated breastfeeding’s impact on certain metabolites and lipids. Conclusions These findings highlight pathways through which breastfeeding reduces inflammation and infection burden, identifying potential targets for optimising infant feeding.