Postoperative cognitive dysfunction (POCD) is a serious complication after surgery, especially in elderly patients. The anesthesia technique is a potentially modifiable risk factor for POCD. This study assessed the effects of dexmedetomidine, propofol or midazolam sedation on POCD in elderly patients who underwent hip or knee replacement under spinal anesthesia. The present study was a prospective randomized controlled preliminary trial. From July 2013 and December 2014, a total of 164 patients aged 65 years or older who underwent hip or knee arthroplasty at China-Japan Friendship Hospital and 41 non-surgical controls were included in this study. Patients were randomized in a 1:1:1 ratio to 3 sedative groups. All the patients received combined spinal-epidural anesthesia (CSEA) with midazolam, dexmedetomidine or propofol sedation. The sedative dose was adjusted to achieve light sedation (bispectral indexBIS score between 70 and 85). All study participants and controls completed a battery of 5 neuropsychological tests before and 7 days after surgery. One year postoperatively, the patients and controls were interviewed over the telephone using the Montreal cognitive assessment 5-minute protocol. In all, 60 of 164 patients (36.6%) were diagnosed with POCD 7 days postoperatively, POCD incidence in propofol group was significantly lower than that in dexmedetomidine and midazolam groups (18.2% vs. 40.0%, 51.9%, χ = 6.342 and 13.603, P = 0.012 and < 0.001). When the patients were re-tested 1 year postoperatively, the incidence of POCD was not significantly different among the 3 groups (14.0%, 10.6% vs. 14.9%, χ = 0.016 and 0.382, P = 0.899 and 0.536). Among dexmedetomidine, propofol and midazolam sedation in elderly patients, propofol sedation shows a significant advantage in term of short-term POCD incidence.This is an open access article distributed under the terms of the Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 (CCBY-NC-ND), where it is permissible to download and share the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal. http://creativecommons.org/licenses/by-nc-nd/4.0.