Background: Because increased serum osmolarity may be lung protective, we hypothesized that increased mortality associated with increased serum sodium would be ameliorated in critically ill patients with an acute respiratory diagnosis. Methods: Data collected within the first 24 hours of intensive care unit (ICU) admission were accessed using ANZICS CORE database. From January 2000 to December 2010, 436 209 patients were assessed. Predefined subgroups including patients with acute respiratory diagnoses were examined. The effect of serum sodium on ICU mortality was assessed with analysis adjusted for illness severity and year of admission. Results are presented as odds ratio (95 confidence interval) referenced against a serum sodium range of 135 to 144.9 mmol/L. Results: Overall ICU mortality was increased at each extreme of dysnatremia (U-shaped relationship). A similar trend was found in various subgroups, with the exception of patients with respiratory diagnoses where ICU mortality was not influenced by high serum sodium (odds ratio, 1.3 [0.7-1.2]) and was different from other patient groups (P <.01). Any adverse associations with hypernatremia in respiratory patients were confined to those with arterial pressure of oxygen (PaO2)/fraction of inspired oxygen (Fio2) ratios of greater than 200. Conclusion: High admission serum sodium is associated with increased odds for ICU death, except in respiratory patients.