PURPOSE: Data on childhood intensive care unit (ICU) deaths are needed to identify changing patterns of intensive care resource utilization. We sought to determine the epidemiology and mode of pediatric ICU deaths in Australia and New Zealand (ANZ). METHODS: This was a retrospective, descriptive study of multicenter data from pediatric and mixed ICUs reported to the ANZ Pediatric Intensive Care Registry and binational Government census. All patients < 16 years admitted to an ICU between 1 January 2006 and 31 December 2016 were included. Primary outcome was ICU mortality. Subject characteristics and trends over time were evaluated. RESULTS: Of 103,367 ICU admissions, there were 2672 (2.6%) deaths, with 87.6% of deaths occurring in specialized pediatric ICUs. The proportion of ANZ childhood deaths occurring in ICU was 12%, increasing by 43% over the study period. Unadjusted (0.1% per year, 95% CI 0.096-0.104; p < 0.001) and risk-adjusted (0.1%/year, 95% CI 0.07-0.13; p < 0.001) ICU mortality rates fell. Across all admission sources and diagnostic groups, mortality declined except following pre-ICU cardiopulmonary arrest where increased mortality was observed. Half of the deaths followed withdrawal of life-sustaining therapy (51%), remaining constant throughout the study. Deaths despite maximal resuscitation declined (0.92%/year, 95% CI 0.89-0.95%; p < 0.001) and brain death diagnoses increased (0.72%/year, 95% CI 0.69-0.75%; p = 0.001). CONCLUSIONS: Unadjusted and risk-adjusted mortality for children admitted to ANZ ICUs is declining. Half of pediatric ICU deaths follow withdrawal of life-sustaining therapy. Epidemiology and mode of pediatric ICU death are changing. Further investigation at an international level will inform benchmarking, resource allocation and training requirements for pediatric critical care.