OBJECTIVES To determine the prevalence of an acquired deficiency of protein S, a coagulation inhibitor, in children infected with the human immunodeficiency virus (HIV) and to identify clinical and laboratory features associated with this coagulation abnormality. METHODS A convenience sample of HIV-infected children, ages 2 to 18 years, was evaluated for total, free and functional protein S; total and functional protein C; prothrombin and activated partial thromboplastin times; fibrinogen; antithrombin III activity; dilute Russell viper venom time; IgG anticardiolipin antibodies; von Willebrand factor antigen; C4b-binding protein; CD4+ T lymphocyte counts; HIV p24 antigen concentration; and serum beta 2-microglobulin concentrations. RESULTS Thirty-four subjects were evaluated. Twenty-four subjects were infected perinatally and 10 by transfusion. Nine of the subjects were CDC Class N (asymptomatic), 13 were Class A/B (symptomatic without AIDS-defining condition) and 12 were Class C (AIDS). None had previously documented thrombosis, nephrosis or significant hepatic dysfunction. Twenty-six subjects (76.5%) had decreased free protein S, and 19 (55.9%) had functional protein S < 2 SD below the mean of laboratory controls. Decreased functional protein S was seen in 33.3% of Class N, 53.8% of Class A/B and 75.0% of Class C subjects. The prevalence of decreased total and functional protein S was greater in those with absolute CD4+ T lymphocyte counts < 200/mm3 compared to those with CD4+ counts > or = 200/mm3 (75.0% vs. 38.9%; chi square, 4.48, P = 0.034). A trend toward negative correlation was observed between protein S and duration of HIV infection only for Class N subjects. No linear correlation was seen between protein S and CD4+ T lymphocyte counts; and no significant relationships were observed between protein S values and CMV status, HIV p24 antigen, C4b-binding protein, von Willebrand factor antigen, IgG anti-cardiolipin antibodies or serum beta 2-microglobulin values. CONCLUSIONS Acquired protein S deficiency is common in HIV-infected children. The high prevalence of this anticoagulant abnormality suggests an increased risk for thrombotic complications in this population.