Rosetting, defined as the binding of two or more uninfected red blood cells (rbc) to an infected rbc, occurs when malarial parasites mature, to trophozoites and schizonts, in the second half of their asexual development. Rosetting is believed to be an important factor in the development of cerebral malaria. In a series of studies to examine the characteristics of the uninfected rbc which contribute to rosetting, the ability of rbc from healthy donors to form rosettes was found to be greater in the cells of group A and B than in those of group O (P = 0.05), and to decrease during storage under blood-blank conditions. Normal rbc exposed for > or = 30 min to quinine, artesunate or artemether (each at 0.25 microgram/ml) in vitro showed significantly decreased rosetting. This effect could not be reversed by extensive washing followed by cultivation for another 24 h in drug-free medium. Mefloquine and pyrimethamine had no effect. Uninfected rbc from patients with uncomplicated or severe falciparum malaria exhibited a lower rosetting ability than rbc from healthy donors (P = 0.01). The rosetting of uninfected rbc of all blood groups from patients with uncomplicated malaria decreased significantly within 2 h of the patients starting treatment with qinghaosu derivatives (artesunate or artemether) and within 8 h of them starting quinine treatment. Similar effects were observed with uninfected rbc from patients with severe malaria after treatment with artesunate but not after quinine. The mechanisms underlying this potentially beneficial effect on rbc adherence are not known.