Inadequate T-cell chimerism following reduced-intensity conditioning transplantation may contribute to graft rejection and disease relapse. Anti-thymocyte globulin (ATG) enhances early donor T-cell chimerism, but may also deplete donor T cells, increasing risks of infection and relapse. We prospectively tested administration of rabbit ATG (rATG) ⩾14 days before the infusion of the graft, followed by in vivo decay of active rATG levels, to selectively deplete host T cells. Twenty-three patients received rATG total dose 4.5 mg/kg on days -16 and -15, fludarabine 30 mg/m(2) per day on day -7 through -3, IV busulfan 130 mg/m(2) per day on days -4 and -3 and cyclophosphamide 1500 mg/m(2) on day -2. rATG levels were therapeutic in all patients on day -14, but were sub-therapeutic (<1 μg/mL) by day 0 in 82% of patients. Median donor T-cell chimerisms on days 30 and 180 were 100% (75-100%) and 100% (90-100%), respectively. Non-relapse mortality and relapse/progression at 48 months were 17 and 30%. Cumulative incidences of acute GvHD grades II-IV and III-IV were 39 and 9%. Median follow-up is 64 months (46-79 months). Survival and disease-free survival at 48 months were 70 and 52%. These data suggest that selective depletion of host T cells using this regimen is a feasible and effective strategy.