Background and objective: High serum osmolarity has been shown to be lung protective. There is lack of clinical studies evaluating the impact on outcomes such as mortality. We aimed to examine the effect of serum osmolarity on intensive care unit (ICU) mortality in critically ill patients. Methods: Data from January 2000 to December 2012 was accessed using the Australian and New Zealand Intensive Care Society (ANZICS) Clinical Outcomes and Resource Evaluation (CORE) database. A total of 509 180 patients were included. Serum osmolarity was calculated from data during the first 24 h of ICU admission. Predefined subgroups (Acute Physiology and Chronic Health Evaluation (APACHE) III diagnostic codes), including patients with acute pulmonary diagnoses, were examined. The effect of serum osmolarity on ICU mortality was assessed with analysis adjusted for illness severity (serum sodium, glucose and urea component removed) and year of admission. Results are presented as OR (95% CI) referenced against a serum osmolarity of 290–295 mmol/L. Results: The ICU mortality was elevated at each extremes of serum osmolarity (U-shaped relationship). A similar relationship was found in various subgroups, with the exception of patients with pulmonary diagnoses in whom ICU mortality was not influenced by high serum osmolarity and was different from other non-pulmonary subgroups (P < 0.01). Any adverse associations with high serum osmolarity in pulmonary patients were confined to patients with a PaO 2 /FiO 2 ratio > 200. Conclusion: High admission serum osmolarity was not associated with increased odds for ICU death in pulmonary patients, unlike other subgroup of patients, and could be a potential area for future interventional therapy.