Polysulfated polysaccharides are attractive candidates for antiviral drug development because of their potent in vitro activities against human immunodeficiency virus (HIV), herpesviruses, and other enveloped viruses. To determine the potential anti-HIV activity of a prototypical polysulfated polysaccharide, we administered the maximally tolerated dose of dextran sulfate by continuous intravenous infusion to 10 subjects with symptomatic HIV infection for up to 14 days. Since parenteral dextran sulfate is an anticoagulant, the infusion was adjusted to produce the greatest acceptable increase in activated partial thromboplastin time. Drug concentrations in plasma achieved with this protocol were up to 200-fold greater than the 50% inhibitory concentration for free HIV infectivity in vitro. Despite this, circulating HIV antigen (p24) levels increased in all eight subjects who received the drug for more than 3 days (median proportional increase, 73.5%; range, 32 to 130%); this increase was highly significant when it was compared with that in a large cohort of untreated historical controls (Fisher's exact test, P less than 0.001). Frequent decreases in infusion rate were required in all subjects to maintain a constant activated partial thromboplastin time; plasma dextran sulfate levels did not fall as the infusion rate decreased, suggesting a decline in estimated drug clearance over time. Continuous intravenous dextran sulfate was toxic, producing profound but reversible thrombocytopenia in all eight subjects who received drug for more than 3 days and extensive but reversible alopecia in five of these subjects. Because of its toxicity and lack of beneficial effect on surrogate markers, dextran sulfate is unlikely to have a practical role in the treatment of symptomatic HIV infection.