Among 1,558 postmenopausal women with non-metastatic breast cancer (mean age 71) from the Cancer Prevention Study-II Nutrition Cohort, cutting back or quitting alcohol did not lower all-cause mortality over 12.6 years of follow-up (750 deaths, only 130 from breast cancer). Quitting between pre- and post-diagnosis assessments was linked to higher mortality (HR 1.39, 95% CI 1.12-1.73), while post-diagnosis cessation showed no association (HR 0.98). Women drinking at least one drink daily who increased intake before diagnosis had higher mortality (HR 1.83, 1.02-3.31). The strongest signal is probably an artifact: the classic 'sick quitter' effect, where women stop drinking because they are already unwell or starting treatment, makes cessation look harmful. That reverse-causation problem plagues observational alcohol research and is not resolved by prospective design alone. Heavy drinking was rare here, so the study cannot say whether sustained reduction helps heavy drinkers, and the two-year change window may be too short to show benefit. The all-cause outcome also dilutes breast-cancer-specific effects, since most deaths were from other causes. The finding does not license drinking after diagnosis; large cohorts link alcohol to higher breast cancer incidence and recurrence risk, and the confidence intervals here are wide enough to hide modest benefit. Treat this as an incremental, confirmatory piece in a muddy literature: it shows how hard late-life alcohol change is to evaluate, not that alcohol is safe for survivors. Survivors should discuss intake with their oncology team.