The relationship between mannitol pharmacokinetics and changes in serum osmolality were studied in dogs and humans. Four human subjects each received between 0.5 and 0.7 g/kg of mannitol as an i.v. infusion given over 15 min. Intravenous bolus doses of 0.5, 1.0 and 1.5 g/kg were given to each of five animals. Serial determinations of serum osmolality and serum mannitol concentrations were then performed. Mannitol disposition was best described using a biexponential equation and assuming a two-compartment, open model with elimination from the central compartment. For human subjects, the mean (+/- S.D.) distribution half-life was 2.11 +/- 2.67 min and the elimination half-life was 71.15 +/- 27.02 min. The volume of distribution was 0.47 +/- 0.50 liters/kg and total body clearance was 7.15 +/- 10.23 ml X min-1 X kg-1. The disposition of mannitol in dogs is similar to that observed in humans. Mannitol clearance was independent of dose whereas the central compartment volume was significantly larger (P less than .005) in animals receiving the 1.5-g/kg dose. The volume change is probably due to a rapid, uncompensated shift of water from intracellular to extracellular space. There was a strong positive correlation (r = 0.90) between mannitol concentration and serum osmolality changes. However, neither maximum serum mannitol concentration nor maximum serum osmolality increased proportionately with dose. Only the 1.5-g/kg dose produced a sustained elevation of serum osmolality, confirming that larger doses are more likely to result in prolonged hypertonic dehydration.