We performed a prospective randomized trial of antithoracic duct lymphocyte globulin (ATDLG), HLA-haploidentical marrow, and androgen (regimen ABA) versus androgen alone (concurrent STANDARD care controls) in 42 newly diagnosed individuals with severe aplastic anemia. ABA patients also were matched with patients from our preceding study (historical STANDARD care controls). Supportive care and pretreatment patient characteristics were the same in all groups. By life table analysis, 76% of patients receiving ABA are alive at 2 yr compared to 31% of the concurrent control group (p less than 0.002 versus ABA) and 19% of the historical controls (p less than 0.0001 versus ABA) given STANDARD care. ABA patients had greater hematologic improvement than either control group (p less than 0.001). However, improvement with ABA was often incomplete. Toxicity of ATDLG was considerable but manageable. Further studies to determine the mechanism of action and active component(s) of ABA are indicated.