Background: Post-operative cognitive dysfunction (POCD) can affect 30% of orthopedic surgery patients. We hypothesized that perioperative temperature has an impact on POCD. Methods: We included 150 patients over 65 years of age scheduled for total knee replacement under spinal anesthesia. They were randomized to receive standard care (sheet cover) or active warming. Neurocognitive assessment (11 subtests) was performed pre-operatively and at day 4 (three subtests) and 3 months (10 subtests). A control group of 55 nonsurgical patients took the same tests at equivalent times. POCD was defined as an individual score decrease of more than 2 standard deviations (SDs) below the baseline on at least two subtests or 2 SDs in the combined z-score, in both cases using control-adjusted changes. Results: Tympanic temperature declined below 35 °C in 88% of standard-care patients; 25.3% of warmed patients had a temperature ≥36 °C. On day 4, 3.2% of standard-care patients and 19.4% of warmed patients had POCD (P=0.0058). At 3 months, there were no between-group differences (standard care, 14.3%; warmed, 6.5%) (P=0.2440). Conclusions: Perioperative warming was associated with a higher incidence of cognitive dysfunction at 4 days after total knee replacement in patients >65 years of age. © 2011 The Acta Anaesthesiologica Scandinavica Foundation.