This retrospective report compared the 4-year outcomes of allogeneic stem cell transplantation (allo-SCT) in 651 adult patients with acute myeloid leukemia receiving a reduced-intensity (RIC) or nonmyeloablative conditioning (NMA) regimen according to the type of unrelated donors. These were either umbilical cord blood (UCB, n = 205), a 9/10 mismatched unrelated donor (MisMUD, n = 99), or a 10/10 matched unrelated donor (MUD, n = 347) graft. Neutrophil recovery was slower in UCB (74.5% by day 42) compared with MisMUD (94.8%) and MUD (95.6%) (P < .001). There was no significant difference in nonrelapse mortality between UCB and both MUD (hazard ratio [HR], 1.05; 95% confidence interval [CI], .62 to 1.78; P = .85) and MisMUD (HR, 1.58; 95% CI, .88 to 2.83; P = .13) The relapse/progression was similar between UCB and MisMUD (HR, .62; 95% CI, .37 to 1.03; P = .07), but was significantly lower in MUD compared with UCB (HR, .60; 95% CI, .39 to .92; P = .02). The rate of extensive chronic graft-versus-host disease (GVHD) was similar between UCB and both MUD (HR, 2.15; 95% CI, .93 to 4.97; P = .08) and MisMUD (HR, 1.84; 95% CI, .68 to 4.95; P = .23). The rate of severe grade III and IV acute GVHD was significantly increased in MisMUD compared with UCB (HR, 2.61; 95% CI, 1.30 to 5.23; P = .007). There was no significant difference in overall survival between UCB and both MisMUD (HR, .98; 95% CI, .66 to 1.45; P = .92) and MUD (HR, .74; 95% CI, .52 to 1.03; P = .08). These data suggest that in the setting of RIC/NMA, allo-SCT UCB is a valid alternative graft source, with significantly less chronic GVHD, compared with MisMUD, when there is no MUD available or when urgent transplantation is needed.