Our objective was to determine how docosahexaenoic acid (DHA) proportions in human milk are modulated by maternal FADS gene variants and dietary intake in Taiwanese women. Inclusion criteria included being healthy, 20–40 y old, having had a full-term baby that they intended to breast feed for at least 1 month, and willingness to participate in this study. Intake of DHA was assessed by food frequency questionnaire and fatty acids were analyzed in human milk samples collected 3–4 weeks postpartum. Based on multiple linear regression of data from 164 mothers that completed this study, there was 0.28% (FA%) reduction in milk DHA in high versus low genetic risk (stratified by whether minor allele numbers were ≥ 3 in rs1535 and rs174448) and 0.45% reduction in low versus high intake (stratified by whether DHA intake reached 200 mg/d). There was a significant gene–diet interaction; mothers with low genetic risk only had high milk DHA proportions with high DHA intake, whereas for mothers with high genetic risk, dietary effects were quite limited. Therefore, for FADS single nucleotide polymorphism in Taiwanese women, increasing DHA intake did not correct low milk DHA proportions in those with a high-risk genotype. Diet only conferred benefits to those with a low-risk genotype. Trial registration: This trial was retrospectively registered (Feb 12, 2019) in ClinicalTrials.gov (No. NCT03842891, https://clinicaltrials.gov/ct2/show/NCT03842891).