A case-control study was conducted at two institutions to determine whether the absolute CD4 lymphocyte count or the percentage of lymphocytes bearing the CD4 marker (i.e., the CD4 percentage) is a more accurate indicator of underlying immune status in splenectomized patients infected with human immunodeficiency virus (HIV). Each of nine splenectomized HIV-infected cases was matched with six nonsplenectomized HIV-infected controls--three matched for CD4 lymphocyte count and three for CD4 percentage. In analyses including the eight cases with an initial CD4 lymphocyte count of > 200/mm3, controlling for the CD4 count revealed differences between cases and controls in terms of CD4 percentage (range, 10%-41% and 17%-54%, respectively; P < .01) and Centers for Disease Control and Prevention (CDC) clinical stage (P = .06). Controlling for the CD4 percentage revealed a significant difference between cases and controls in terms of CD4 count (range, 396-1,040 and 55-784 cells/mm3, respectively; P < .01) but not CDC clinical stage (P > .7). These data suggest that the numerical relationship between the CD4 lymphocyte count and the CD4 percentage among splenectomized HIV-infected patients with more than 200 CD4 cells/mm3 differs from that among nonsplenectomized patients. The CD4 percentage appears to be a more accurate indicator of the underlying level of immune function in the former group of patients.