Background:　Alcohol　use　impacts　brain　structure,　including　white　matter　integrity,　which　can　be　quantified　by　fractional　anisotropy　(FA)　in　diffusion　tensor　imaging　(DTI).　This　study　explored　the　relationship　between　the　severity　of　alcohol　consumption　and　white　matter　FA　changes,　and　its　sex　differences,　in　young　adults,　using　data　from　the　Human　Connectome　Project.　Methods:　We　analyzed　DTI　data　from　949　participants　(491　females)　and　used　principal　component　analysis　(PCA)　of　15　drinking　metrics　to　quantify　drinking　severity.　Connectome-based　predictive　modeling　(CPM)　was　employed　to　predict　the　principal　component　of　drinking　severity　from　network　FA　values　in　a　matrix　of　116×116　regions.　Mediation　analyses　were　conducted　to　explore　the　interrelationships　among　networks　identified　by　CPM,　drinking　severity,　and　rule-breaking　behavior.　Results:　Significant　correlations　were　found　between　drinking　severity　and　network　FA　values.　Both　men　and　women　showed　significant　correlations　between　negative　network　connectivity　and　drinking　severity　(men:　r=0.15,　P=0.001;　women:　r=0.30,　P＜0.001).　Sex　differences　were　observed　in　the　brain　regions　contributing　to　drinking　severity　predictions.　Mediation　analyses　revealed　significant　inter-relationships　between　network　features,　drinking　severity,　and　rule-breaking　behavior.　Conclusions:　The　connectomics　of　white　matter　FA　can　predict　the　severity　of　alcohol　consumption,　and　by　incorporating　brain　network　pathways,　identify　sex　differences.　This　approach　provides　new　clues　to　the　biological　basis　of　alcohol　abuse　and　evaluates　how　these　regions　interact　in　broader　brain　networks　for　understanding　alcohol　misuse　and　its　comorbidities.　©　AME　Publishing　Company.