Ready-to-use supplemental foods (RUSF) are energy-dense meals formulated to prevent and treat moderate and severe childhood acute malnutrition (MAM and SAM) in high-risk settings. Although lifesaving, the degree and durability of weight recovery with RUSF is unpredictable. We examined whether environmental enteric dysfunction (EED) and gut microbiota perturbations are risk factors for RUSF failure in a birth cohort of 416 rural Pakistani children followed for growth, common childhood illnesses, and biomarkers from blood, urine, and stool. Infants who developed wasting (weight-for-length Z score <-2, n=187, 45%) during surveillance received Acha Mum (a chickpea-based RUSF) daily for eight weeks. Machine learning identified seven biomarkers that predicted RUSF response (n=75) vs. non-response (n=112) with 73% accuracy. Remarkably, gut microbiome composition predicted RUSF response with 93% (pre-supplementation) and 98% (post-supplementation) accuracy. Seven outliers whose microbiome falsely predicted positive response experienced extraordinary burdens of inflammation and illness during supplementation. These findings identify gut microbial signatures and biomarkers of gut and systemic inflammation as robust predictors of RUSF response in infants free from intercurrent illness during recovery, setting the stage for predictive models to guide precision use of RUSF and adjunct therapies in undernourished children.