Serum selenium levels were determined cross-sectionally in 57 HIV-infected patients who were classified according to the Centers for Disease Control (CDC) 1993 classification system. Mean serum selenium levels were lower in CDC stage II (58.7 +/- 12.2 micrograms/L; p < 0.01; n = 18) and stage III (47.6 +/- 11.3 micrograms/L; p < 0.01; n = 19) HIV-infected patients, than in healthy subjects (80.6 +/- 9.6 micrograms/L; n = 48) and stage I patients (73.6 +/- 16.5 micrograms/L; n = 20). Serum selenium levels were positively correlated with CD4 count, CD4/8 ratio, hematocrit, and serum albumin (r = 0.42; r = 0.39; r = 0.48; and r = 0.45; p < 0.01, respectively) and inversely with serum levels of thymidine kinase (r = -0.49; p < 0.01; n = 49) and beta 2-microglobulin (r = -0.46; p < 0.001; n = 49). In addition, serum selenium levels in 20 randomly selected AIDS-free individuals (CDC I: n = 10; CDC II: n = 10) were inversely correlated with serum concentrations of interleukin-8 (IL-8) and soluble tumor necrosis factor receptors (sTNFR) types I and II. There was no correlation with serum immuneglobulin A and total serum protein levels. The results show that the progressive deprivation of serum selenium in HIV-infection is associated with loss of CD(4+)-cells and with increased levels of markers of disease progression and inflammatory response.