Rodents given cyclosporhe (CSP) for several weeks after autologous or syngeneic bone marrow transplantation develop a syndrome that mimics allogeneic graft-versus-host disease (GVHD). Autologous GVHD has also been reported after administration of CSP in patients who have received autologous bone marrow transplantation (ABMT) with untreated marrow for lym-phoma or acute myeloid leukemia (AML). Our study was designed to determine whether CSP administration is associated with appearance of autologous GVHD in patients with AML receiving ABMT with 4-hydroperoxycyclophosphamide (4HC)-purged marrow and whether there was a dose-dependent effect of CSP on development of the syndrome. Thirty-three patients with AML (18 in first remission [CRl], 10 in CR2, and 5 in CR3) received intravenous CSP, beginning on the day of ABMT, after a preparative regimen of busulfan and cyclophosphamide and ABMT with 4HC-treated marrow. Skin biopsies were obtained weekly after ABMT or on appearance of rash and were graded for GVH changes. In the first phase of this study, groups of patients received CSP dosages of either 1 mg/kg/day (7 patients), 2.5 mg/kg/day (8 patients), or 3.75 mg/kg/day (6 patients) for 28 days. Sixteen of the 21 patients (76%) developed cutaneous histopathologic grade 2 GVHD at a median of 34 days (range, 14-49) after ABMT, and cutaneous manifestations were present at time of positive biopsy in 11 of the 16 patients. There was no apparent difference in frequency, time to onset, or duration of GVHD among the three CSP dosage groups. In the subsequent trial, 12 patients received 1.0 mg/kg/day of CSP for 35 days; 6 developed biopsy-documented GVHD at a median of 42 days (range, 18-52) after ABMT. Overall, 22 of 33 patients (67%) had positive biopsies for GVHD, compared with a historical 7% incidence of spontaneous GVHD in ABMT recipients. No patients had hepatic or gastrointestinal dysfunction attributable to GVHD or required specific therapy for GVHD. Six of the 33 patients died with ABMT-related complications; 3 had positive biopsies for cutaneous GVHD. Eleven patients (6 CR1, 4 CR2, 1 CR3) relapsed with AML at a median of 285 days (range, 115-633) after ABMT; 9 had positive biopsies. Sixteen patients (9 CRl, 3 CR2, 4 CR3) are alive without relapse at a median of 509+days (range, 96+-1176+) after ABMT; 10 had cutaneous GVHD. Randomized prospective trials will be needed to determine whether autologous GVHD is associated with alterations in relapse rate and disease-free survival after ABMT for acute leukemia and lymphoma.