Background: The pattern of cortisol secretion is influenced by surgery. As cortisol can adversely affect neuronal function, this may be an important factor in the development of post-operative cognitive dysfunction (POCD). We hypothesized that the incidence of POCD would be related to changes in cortisol level. Methods: We studied 187 patients aged over 60 years undergoing major non-cardiac surgery with general or regional anaesthesia. Saliva cortisol levels were measured pre-operatively and at 1 day, 7 days and 3 months post-operatively in the morning (08.00 h) and in the afternoon (16.00 h) using salivettes. Cognitive function was assessed pre-operatively, on day 7 and at 3 months using four neuropsychological tests. POCD was defined as a combined Z score of greater than 1.96. Results: After surgery, salivary cortisol concentrations increased significantly. POCD was detected in 18.8% of subjects at 1 week and in 15.2% after 3 months. The pre-operative ratios between the morning and afternoon cortisol concentrations (am/pm ratios) were 2.8 and 2.7 in patients with POCD at 1 week vs. those without POCD at 1 week, respectively. The am/pm ratios decreased significantly post-operatively to 1.9 and 1.6 at 1 week, respectively (P = 0.02 for both). In an analysis considering all am/pm ratios, it was found that the persistent flattening in am/pm ratio was significantly related to POCD at 1 week. Conclusion: The pattern of diurnal variation in cortisol level was significantly related to POCD. Thus, circadian rhythm disturbance or metabolic endocrine stress could be an important mechanism in the development of cognitive dysfunction after major surgery.