Delirium is a common and disturbing postoperative complication that might be ameliorated by propofol-based anaesthesia. We therefore tested the primary hypothesis that there is less delirium after propofol-based than after sevoflurane-based anaesthesia within 7 days of major cancer surgery. This multicentre randomised trial was conducted in 14 tertiary care hospitals in China. Patients aged 65–90 yr undergoing major cancer surgery were randomised to either propofol-based anaesthesia or to sevoflurane-based anaesthesia. The primary endpoint was the incidence of delirium within 7 postoperative days. A total of 1228 subjects were enrolled and randomised, with 1195 subjects included in the modified intention-to-treat analysis (mean age 71 yr; 422 35% women); one subject died before delirium assessment. Delirium occurred in 8.4% (50/597) of subjects given propofol-based anaesthesia vs 12.4% (74/597) of subjects given sevoflurane-based anaesthesia (relative risk 0.68 95% confidence interval {CI}: 0.48–0.95; P=0.023; adjusted relative risk 0.59 95% CI: 0.39–0.90; P=0.014). Delirium reduction mainly occurred on the first day after surgery, with a prevalence of 5.4% (32/597) with propofol anaesthesia vs 10.7% (64/597) with sevoflurane anaesthesia (relative risk 0.50 95% CI: 0.33–0.75; P=0.001). Secondary endpoints, including ICU admission, postoperative duration of hospitalisation, major complications within 30 days, cognitive function at 30 days and 3 yr, and safety outcomes, did not differ significantly between groups. Delirium was a third less common after propofol than sevoflurane anaesthesia in older patients having major cancer surgery. Clinicians might therefore reasonably select propofol-based anaesthesia in patients at high risk of postoperative delirium. Chinese Clinical Trial Registry (ChiCTR-IPR-15006209) and ClinicalTrials.gov (NCT02662257).