We studied the effect of the CD34+ cell dose on transplant-related mortality (TRM) and survival in 39 patients randomized to receive lenograstim-mobilized PBSCT (n = 20) or BMT (n = 19) from HLA-identical siblings. Both marrow and blood were harvested, and one infused in a double-blind fashion. The median nucleated (7.0 vs 3.2 x 10(8)/kg; P < 0.0001), CD34+ (3.7 vs 1.5 x 10(6)/kg; P = 0.002), CFU-GM (42 vs 19 x 10(4)/kg; P = 0.002), and CD3+ (1.9 vs 0.3 x 10(8)/kg; P < 0.0001) cell doses with PBSCT were higher. Thirteen patients (6 BMT and 7 PBSCT) experienced TRM at 15-733 days (median 57); 10 of 20 receiving <2 x 10(6) CD34+ cells/kg compared with three of 19 receiving > or =2. Eight of 20 patients receiving <2 x 10(6) CD34+ cells/kg are alive compared with 14 of 19 receiving > or =2. In Cox analysis, CD34+ cell dose > or =2 x 10(6)/kg was associated with lower TRM (RR 0.2, P = 0.01), and higher overall (RR 3.7, P = 0.01) and event-free (RR 3.2, P = 0.02) survival. Other cell populations and the source of stem cells did not affect TRM or survival. We conclude that 2 x 10(6) CD34+ cells/kg may be the ideal minimum cell dose for allogeneic transplantation although lower doses do not preclude successful therapy. Since the likelihood of obtaining this threshold CD34+ cell number is significantly greater from blood than marrow, PBSCT may be preferable to marrow for allografts from HLA-identical siblings.