The reference method to measure serum osmolality (Mosm) is the delta cryoscopic method. However, the technology may not be available. Clinicians therefore must calculate osmolality (Cosm) as the sum of concentrations of selected principal solutes such as sodium, potassium, urea, and glucose. To evaluate the validity of Cosm in patients in hyperosmolar state we compared Cosm to Mosm. Twenty-two ICU patients treated by infusion of hypertonic solutes for intracranial hypertension following head injury were prospectively studied. A control group of 10 patients with hypernatremia due to medical causes was also evaluated. Na+, blood urea nitrogen (BUN), and glucose were measured to calculate serum osmolality: Cosm = (2 x Na+) + BUN + glucose (in mOsm/kg). Measurement of serum osmolality was performed using the delta-cryoscopic method. The results of the two methods were compared by correlating the difference (Mosm-Cosm) between each pair of results with the mean of the pairs of results. Cosm underestimated Mosm (3.4 mOsm/kg, P < 0.02) in the control group whereas Cosm overestimated Mosm in patients who received hypertonic fluids (2.3 mOsm/kg, P < 0.03). Calculation of osmolality introduced a systematic bias, overestimating osmolality in the lower ranges and underestimating it in the higher ranges. In the range of hyperosmolality commonly used to manage intracranial hypertension following head trauma, serum osmolality, as determined by sum of principal serum solutes, compares poorly with direct measurement using standard osmometry.