The incidence and impact of postoperative complications are poorly described. Failure-to-rescue, the rate of death following complications, is an important quality measure for perioperative care but has not been investigated across multiple health care systems. Methods. We analysed data collected during the International Surgical Outcomes Study, an international 7-day cohort study of adults undergoing elective inpatient surgery. Hospitals were ranked by quintiles according to surgical procedural volume (Q1 lowest to Q5 highest). For each quintile we assessed in-hospital complications rates, mortality, and failure-to-rescue. We repeated this analysis ranking hospitals by risk-adjusted complication rates (Q1 lowest to Q5 highest). Results. A total of 44 814 patients from 474 hospitals in 27 low-, middle-, and high-income countries were available for analysis. Of these, 7508 (17%) developed one or more postoperative complication, with 207 deaths in hospital (0.5%), giving an overall failure-to-rescue rate of 2.8%. When hospitals were ranked in quintiles by procedural volume, we identified a threefold variation in mortality (Q1: 0.6% vs Q5: 0.2%) and a two-fold variation in failure-to-rescue (Q1: 3.6% vs Q5: 1.7%). Ranking hospitals in quintiles by risk-adjusted complication rate further confirmed the presence of important variations in failureto- rescue, indicating differences between hospitals in the risk of death among patients after they develop complications. Conclusions. Comparison of failure-to-rescue rates across health care systems suggests the presence of preventable postoperative deaths. Using such metrics, developing nations could benefit from a data-driven approach to quality improvement, which has proved effective in high-income countries.