We analyzed pretransplant donor anti-recipient cytotoxic and helper T lymphocyte (CTL and HTL) responses separately in two cohorts of unrelated marrow transplant recipients. Donors and recipients were typed for HLA-A and -B antigens by serologic methods, and for HLA-DRB1 by molecular methods. A single mismatch for a cross-reactive HLA-A or -B antigen or the -DRB1 allele was accepted in patients younger than 36 years if an HLA-A, -B, or -DRB1-matched donor could not be identified. The combination of methotrexate and cyclosporine was used for graft-vs.-host disease (GVHD) prophylaxis, and marrows were not T cell depleted. Donor anti-recipient CTL precursor frequencies showed no correlation with the severity of acute GVHD or with survival after transplantation. HTL responses were detected in the presence of HLA-class II disparity and showed weak correlations with the severity of acute GVHD (p = 0.054) and with survival after transplantation (p = 0.08). These results suggest that testing donor anti-recipient CTL responses before unmodified marrow transplantation does not predict clinically important events and is not likely to help select unrelated donors. With the current availability of molecular genetic methods for assessing HLA-class II compatibility, testing donor anti-recipient HTL responses is not likely to add information that would help select unrelated donors.