Purpose T-cell-replete HLA-haploidentical donor hematopoietic transplantation using post-transplant cyclophosphamide was originally described using bone marrow (BM). With increasing use of mobilized peripheral blood (PB), we compared transplant outcomes after PB and BM transplants. Patients and Methods A total of 681 patients with hematologic malignancy who underwent transplantation in the United States between 2009 and 2014 received BM (n = 481) or PB (n = 190) grafts. Cox regression models were built to examine differences in transplant outcomes by graft type, adjusting for patient, disease, and transplant characteristics. Results Hematopoietic recovery was similar after transplantation of BM and PB (28-day neutrophil recovery, 88% v 93%, P = .07; 100-day platelet recovery, 88% v 85%, P = .33). Risks of grade 2 to 4 acute (hazard ratio [HR], 0.45; P < .001) and chronic (HR, 0.35; P < .001) graft-versus-host disease were lower with transplantation of BM compared with PB. There were no significant differences in overall survival by graft type (HR, 0.99; P = .98), with rates of 54% and 57% at 2 years after transplantation of BM and PB, respectively. There were no differences in nonrelapse mortality risks (HR, 0.92; P = .74) but relapse risks were higher after transplantation of BM (HR, 1.49; P = .009). Additional exploration confirmed that the higher relapse risks after transplantation of BM were limited to patients with leukemia (HR, 1.73; P = .002) and not lymphoma (HR, 0.87; P = .64). Conclusion PB and BM grafts are suitable for haploidentical transplantation with the post-transplant cyclophosphamide approach but with differing patterns of treatment failure. Although, to our knowledge, this is the most comprehensive comparison, these findings must be validated in a randomized prospective comparison with adequate follow-up.