Background: Postoperative cognitive dysfunction (POCD) is a common complication after cardiac and major non‐cardiac surgery with general anaesthesia in the elderly. We hypothesized that the incidence of POCD would be less with regional anaesthesia rather than general. Methods: We included patients aged over 60 years undergoing major non‐cardiac surgery. After giving written informed consent, patients were randomly allocated to general or regional anaesthesia. Cognitive function was assessed using four neuropsychological tests undertaken preoperatively and at 7 days and 3 months postoperatively. POCD was defined as a combined Z score >1.96 or a Z score >1.96 in two or more test parameters. Results: At 7 days, POCD was found in 37/188 patients (19.7%, [14.3–26.1%]) after general anaesthesia and in 22/176 (12.5%, [8.0–18.3%]) after regional anaesthesia, P = 0.06. After 3 months, POCD was present in 25/175 patients (14.3%, [9.5–20.4%]) after general anaesthesia vs. 23/165 (13.9%, [9.0–20.2%]) after regional anaesthesia, P = 0.93. The incidence of POCD after 1 week was significantly greater after general anaesthesia when we excluded patients who did not receive the allocated anaesthetic: 33/156 (21.2%[15.0–28.4%]) vs. 20/158 (12.7%[7.9–18.9%]) (P = 0.04). Mortality was significantly greater after general anaesthesia (4/217 vs. 0/211 (P < 0.05)). Conclusion: No significant difference was found in the incidence of cognitive dysfunction 3 months after either general or regional anaesthesia in elderly patients. Thus, there seems to be no causative relationship between general anaesthesia and long‐term POCD. Regional anaesthesia may decrease mortality and the incidence of POCD early after surgery.