We have studied the ability of heparin to disrupt spontaneous rosettes formed between Plasmodium falciparum-infected and uninfected red blood cells, which has been proposed to have importance in the pathogenesis of cerebral malaria. Substantial variation in this activity was found among six laboratory stains of P. falciparum. Rosettes formed by three of these strains were highly sensitive to heparin (50% disruption at 0.5-25 micrograms/ml; 1 microgram/ml corresponds to 0.15 IU/ml). The rosettes formed by two other strains showed a much lower sensitivity (50% disruption at 700-2,500 micrograms/ml), while the rosettes formed by another strain were almost completely resistant to heparin (20% disruption at 6,500 micrograms/ml). The ability of heparin (65 or 650 micrograms/ml) to disrupt rosettes formed by 54 fresh Gambian isolates of P. falciparum also varied. Rosettes of 27 (50%) of the 54 isolates were disrupted to a significant degree (greater than or equal to 15%), while rosettes of the other 27 isolates remained unaffected at the concentrations tested. Heparin was fractionated by molecular weight and/or affinity for antithrombin III. We found that its property of rosette disruption was associated, to some extent, with size (high molecular weight) but not with its anticoagulant potential (affinity for antithrombin III). A heparin fraction with low affinity for antithrombin III and one with combined high molecular weight and low affinity for antithrombin III were as effective at disrupting rosettes as standard heparin, while a chemically modified (N-acetylated) high molecular weight-heparin fraction, similarly devoid of anticoagulant activity, lacked strong anti-rosette potential.(ABSTRACT TRUNCATED AT 250 WORDS)