BACKGROUND: Two trials reported that a high inspiratory oxygen fraction (Fio2) does not promote myocardial infarction or death. Observational studies can provide larger statistical strength, but associations can be due to unobserved confounding. Therefore, we evaluated the association between intraoperative Fio2 and cardiovascular complications in a large international cohort study to see if spurious associations were observed. METHODS: We included patients from the Vascular events In noncardiac Surgery patIents cOhort evaluatioN (VISION) study, who were ≥45 years of age, scheduled for overnight hospital admission, and had intraoperative Fio2 recorded. The primary outcome was myocardial injury after noncardiac surgery (MINS), and secondary outcomes included mortality and pneumonia, all within 30 postoperative days. Data were analyzed with logistic regression, adjusted for many baseline cardiovascular risk factors, and illustrated in relation to findings from 2 recent controlled trials. RESULTS: We included 6588 patients with mean age of 62 years of whom 49% had hypertension. The median intraoperative Fio2 was 0.46 (5%–95% range, 0.32–0.94). There were 808 patients (12%) with MINS. Each 0.10 increase in median Fio2 was associated with a confounder-adjusted increase in odds for MINS: odds ratio (OR), 1.17 (95% confidence interval [CI], 1.12–1.23; P < .0001). MINS occurred in contrast with similar frequencies and no significant difference in controlled trials (2240 patients, 194 events), in which patients were given 80% vs 30% oxygen. Mortality was 2.4% and was not significantly associated with a median Fio2 (OR, 1.07; 95% CI, 0.97–1.19 per 0.10 increase; P = .18), and 2.9% of patients had pneumonia (OR, 1.05; 95% CI, 0.95–1.15 per 0.10 increase; P = .34). CONCLUSIONS: We observed an association between intraoperative Fio2 and risk of myocardial injury within 30 days after noncardiac surgery, which contrasts with recent controlled clinical trials. Fio2 was not significantly associated with mortality or pneumonia. Unobserved confounding presumably contributed to the observed association between Fio2 and myocardial injury that is not supported by trials.