We describe a novel animal model of nonmyeloablative bone marrow transplantation (BMT) using the purine analog pentostatin. Other cohorts of mice received another purine analog, fludarabine, which we and others have previously evaluated in nonmyeloablative murine models. We evaluated pentostatin for its ability to (1) operate synergistically with cyclophosphamide to induce host T cell depletion; (2) induce host T cell suppression, as defined by modulation of cytokine secretion in vitro and abrogation of host-versus-graft reactivity in vivo; (3) constrain host T cell recovery post-therapy; and (4) prevent the rejection of T cell–depleted, fully major histocompatibility complex–mismatched bone marrow allografts. Relative to single-agent regimens, combination regimens with pentostatin and cyclophosphamide (PC) and with fludarabine and cyclophosphamide (FC) worked synergistically to deplete host CD4⁺ and CD8⁺ T cells. PC and FC regimens were developed that yielded similar levels of host T cell and myeloid cell depletion. In the setting of these generally comparable states of host T cell and myeloid cell depletion, the PC regimen was found to be highly immunosuppressive, as evidenced by a reduced host T cell capacity to secrete interleukin-2 and interferon-γ in vitro, to mediate host-versus-graft reactivity in vivo, and to recover numerically and functionally during a 2-week observation period after chemotherapy. Finally, using B6 hosts treated with the 14-day chemotherapy regimens, the PC regimen more consistently prevented the rejection of BALB/c T cell–depleted allografts compared with the FC regimen (rate of alloengraftment, 14/15 [93%] of PC-treated recipients vs 8/14 [57%] of FC-treated recipients; P < .05); similar results were observed using an 8-day conditioning regimen. These data suggest that host T cell suppression, distinct from T cell depletion, may be a critical determinant of engraftment after purine analog–based regimens and also may be preferentially attained by the use of pentostatin.