The purpose of this study was to investigate the efficacy and feasibility of unmanipulated haploidentical PBSCT for the treatment of acute leukemia (AL). This study compares the clinical outcomes of high-risk AL patients who received PBSCs harvested from family members sharing at least one common haplotype to outcomes of high-risk AL patients who received a mixture of G-CSF-primed BM (G-BM) and peripheral blood (G-PB) harvests. The results show that PBSCT achieved inferior cumulative myeloid engraftment at 30 days after transplant (89.9 + or - 10.1% vs 100%; P=0.04), with lower cumulative incidence of grade II-IV acute GVHD (aGVHD) (37.1 + or - 16.5% vs 63.2+/-6%; P=0.058) compared with G-BM/G-PB transplant. However, both transplant protocols had similar rates of 2-year relapse (29.6 + or - 17.1% vs 34.0 + or - 5.7%; P=0.954), and PBSCT produced a higher incidence of 2-year non-leukemic mortality (62.5 + or - 14.8% vs 35.1 + or - 5.1%; P=0.014), as well as lower rates of overall (26.8 + or - 12.3% vs 43.2 + or - 5.0%; P=0.052) and disease-free survival (26.8 + or - 12.3% vs 42.4 + or - 5.0%; P=0.071) compared with G-BM/G-PB transplant. These results suggest that haploidentical HSCT is an option for patients with AL who urgently need a graft and do not have matched sibling donors. PBSCT is potentially inferior to G-BM/G-PB transplant, and improvements should be made before PBSCT becomes a routine in unmanipulated mismatched/haploidentical transplant settings.