OBJECTIVE: As hydratation of the normal brain is much more dictated by osmotic gradients than by hydrostatic or oncotic pressures, this study aimed to compare the effect of the infusion of currently used volume loading solutions on plasma osmolality. STUDY DESIGN: Randomized, comparative trial. PATIENTS: Thirty ASA 1-2 patients, scheduled for lumbar intervertebral disc surgery were randomly allocated to three groups receiving either 2,000 mL of lactated Ringer's solution (RL, n = 10), 750 mL of hydroxyethylstarch 6% (HEA, n = 10) or 2,000 mL of normal saline (NaCl, n = 10). METHODS: Baseline osmolality, natraemia, glycaemia and protidaemia were measured before induction of anaesthesia (T1), after the infusion of 375 mL of hydroxyethylstarch or 1,000 mL of crystalloids (T2) and at the end of the infusion (T3). RESULTS: The three groups were identical for age, weight, initial plasma osmolality and natraemia. However, osmolality in the RL group was decreased at T2 and T3 compared to T1 (respectively: 299 +/- 5 mOsm.kg-1, 295 +/- 4 mOsm.kg-1 and 292 +/- 5 mOsm.kg-1. Osmolality at T2 and T3 was also lower in the RL group compared to the HEA and NaCl groups (respectively: 301 +/- 6 mOsm.kg-1 and 304 +/- 13 mOsm.kg-1 for T2 and T3 in the HEA group, and 299 +/- 5 mOsm.kg-1 and 298 +/- 5 mOsm.kg-1 in the NaCl group). In the HEA and NaCl groups, osmolality was unchanged at T2 and T3 compared to T1. CONCLUSION: Both normal saline and hydroxyethylstarch 6% maintain plasma osmolality, whereas Ringer lactate tends to decrease it. For that reason normal saline and hetastarch 6% but not lactated Ringer's solution, may be administered in patients experiencing blood-brain barrier damage.