Both in children and adults, acute leukemia may present with extremely high blast counts; a phenomenon known as hyperleukocytosis. Respiratory failure, intracranial bleeding, and severe metabolic abnormalities frequently occur in acute hyperleukocytic leukemias (AHLs) and are the primary determinants of the high early mortality (20% to 40%) observed. The process leading to these complications has long been known as leukostasis, but the biological mechanisms underlying its development and progression have remained unclear. Traditionally, leukostasis has been attributed to overcrowding of leukemic blasts in the microcirculation, and its treatment has focused on prompt leukocytoreduction. However, it is becoming increasingly evident that leukostasis results from the adhesive interactions between leukemic blasts and the endothelium; a mechanism that none of the current therapies directly addresses. The endothelial damage associated with leukostasis is likely to be mediated by cytokines released in situ and by subsequent migration of leukemic blasts in the perivascular space. The adhesion molecules displayed by the leukemic blasts and their chemotactic response to the cytokines in the vascular microenvironment are probably more important in causing leukostasis than the cell number. This may explain why leukostasis may develop in some patients with AHL and not in others, and why some patients with acute leukemia without hyperleukocytosis (<50,000 blasts/mm(3)) develop leukostasis and respond to leukocytoreduction. Leukapheresis effectively reduces the blast count in many patients with AHL and is routinely used for immediate leukocytoreduction. However, the most appropriate use of leukapheresis in acute leukemia remains unclear, and the procedure may not prevent early death more efficiently than fluid therapy, hydroxyurea, and prompt induction chemotherapy. The use of cranial irradiation remains very controversial and is not generally recommended. The identification of the adhesion molecules, soluble cytokines, and chemotactic ligand-receptor pairs mediating endothelial cell damage in AHL should become a priority if better outcomes are desired.