BACKGROUND Pasteurized human donor milk (DM) is recommended by the World Health Organization and American Academy of Pediatrics for preterm infants when mother's own milk is unavailable, yet the extent and predictors of use and criteria for use in US neonatal intensive care units (NICUs) are unknown. OBJECTIVE This study aimed to evaluate current DM use in US level 3 NICUs and predictors and criteria of use. METHODS We sent mail surveys to 302 US level 3 NICU directors. We used multivariable logistic regression to analyze predictors of DM use. RESULTS Survey response rate was 60%, and 76 of 182 (42%) directors reported DM use. Among DM users, 30% have used DM < 2 years and 55% for 2 to 5 years. Among nonusers, 63% were uncertain of turnaround time when ordering DM, 36% were unclear what guidelines milk banks followed, and 31% were unsure of parent receptiveness. In multivariate analyses, > 800 annual admissions (odds ratio [OR], 4.11; 95% confidence interval [CI], 1.43-11.82; reference ≤ 400 admissions) and location in the Midwest (OR, 3.02; 95% CI, 1.17-7.76) and West (OR, 6.33; 95% CI, 2.28-15.57; reference Northeast) were positively associated with DM use; safety-net hospitals (> 75% Medicaid insurance) were negatively associated (OR, 0.35; 95% CI, 0.14-0.89). CONCLUSIONS Pasteurized human donor milk use is rapidly emerging among US level 3 NICUs. Larger NICUs and those in the West and Midwest were more likely to use DM, while safety-net hospitals were less likely to use DM. Lack of knowledge by medical directors of accessibility, safety, and parental receptiveness may be barriers to DM use.