Background: Dysnatremia, which is associated with increased mortality in general intensive care units (ICU), has not been thoroughly studied in neurologic ICU (NICU). Methods: Prevalence of dysnatremia was retrospectively assessed. The multivariable binary logistic regression model was used to determine the influence of dysnatremia on mortality. Results: Of 519 patients, 106 (20.4%) were admitted with hyponatremia and 177 (34.10%) with hypernatremia. Hypernatremia was detected in 69 (13.29%) patients on admission to NICU and in 108 patients (20.81%) during the ICU stay. However, the incidence of dysnatremia did not differ across the neurological categories (p = 0.4690). ICU stay in patients with acquired hypernatremia (22.3 +/- 25.35 days) was longer than those with admission hypernatremia (13.5 +/- 12.8 days) or with consistent normonatremia (16.16 +/- 20.06 days). The other indicators such as Acute Physiology and Chronic Health Evaluation II, Glasgow Coma Scale score, urinary catheterization, and incidence of pneumonia were also associated with the serum sodium concentrations. Hypernatremia both on admission and acquired in NICU could significantly differentiate between survivors and nonsurvivors (p = 0.002 and <0.0001). However, only NICU-acquired hypernatremia was the independent risk factor for mortality with high sensitivity (p = 0.000). Conclusions: Dysnatremia is more common in NICU, whereas only acquired-hypernatremia was independently associated with outcome. (C) 2014 S. Karger AG, Basel