We studied 61 patients with a closed head injury and increased intracranial pressure (ICP). The ICP was monitored continuously, concomitant with the administration of 20% mannitol. If the ICP remained higher than 25 mm Hg for 10 minutes or more, the patient was included in the study. Analysis of monitoring records delineated four variables that were related to the response of ICP to mannitol: (a) the level of ICP 1 hour before mannitol was administered, (b) the level of ICP when mannitol was administered, (c) the amount of mannitol that was administered immediately before the resulting changes in ICP were measured, and (d) the cumulative amount of mannitol given over the 6 hours before the most recent mannitol dosage was administered. The level of the ICP measurements and the cumulative amount of preceding doses of mannitol influenced the response of ICP to mannitol more than did the size of the dose of mannitol. These findings imply that (a) the initial administration of more mannitol than is absolutely needed may lead to larger doses being required to control ICP and (b) for that reason, mannitol given on a gram/kilogram, an hourly, or a serum osmolarity basis to control increased ICP has negative long term effects because more mannitol may be required to decrease ICP when an excessive amount of it has been given previously.