The pathogenesis of brain swelling and neurological deterioration after rapid hemodialysis (dialysis disequilibrium syndrome) is controversial. The "reverse urea hypothesis" suggests that hemodialysis removes urea more slowly from the brain than from the plasma, creating an osmotic gradient that results in cerebral edema. The "idiogenic osmole hypothesis" proposes that an osmotic gradient between brain and plasma develops during rapid dialysis because of newly formed brain osmoles. In this review, the experimental basis for the two hypotheses are critically examined. Based on what is known about the physiology of urea and water diffusion across the blood-brain barrier, and empiric observations of brain solute composition after experimental hemodialysis, we conclude that the "reverse urea hypothesis" remains a viable explanation for dialysis disequilibrium and that rapid reduction of a high urea level in and of itself predisposes to this condition.