Objectives Folate is an essential nutrient fundamental to human growth and development. Human milk maintains high folate content across the maternal folate status range, suggesting buffering of milk folate with prioritized delivery to milk at the expense of maternal depletion. We investigated whether and how the extent of this buffering may diminish under prolonged nutritional and/or disease stress, while taking into consideration infants' varying vulnerability to malnutrition-related morbidity/mortality. Methods A cross-sectional study analyzed milk specimens from northern Kenyan mothers (n = 203), surveyed during a historic drought and ensuing food shortage. Multiple regression models for folate receptor-alpha (FOLR1) in milk were constructed. Predictors included maternal underweight (BMI < 18.5), iron-deficiency anemia (hemoglobin 5 mg/L), folate deficiency (hyperhomocysteinemia, homocysteine >12 or 14 mu mol/L), inflammation (serum C-reactive protein >5 mg/L), infant age and sex, and mother-infant interactions. Results In adjusted models, milk FOLR1 was unassociated with maternal underweight, iron-deficiency anemia and inflammation. FOLR1 was positively associated with maternal folate deficiency, and inversely associated with infant age. There was interaction between infant age and maternal underweight, and between infant sex and maternal folate deficiency, predicting complex changes in FOLR1. Conclusions Our results suggest that mothers buffer milk folate against their own nutritional stress even during a prolonged drought; however, the extent of this buffering may vary with infant age, and, among folate-deficient mothers, with infant sex. Future research is needed to better understand this variability in maternal buffering of milk folate and how it relates to folate status in nursing infants.