Background Robot-assisted total knee arthroplasty (TKA) has been largely studied to confirm its advantages in terms of accurate component positioning, microembolus formation, less blood loss, and so on, but is currently usually performed under tourniquet due to its longer operative time than conventional TKA. The aim of this study was to estimate the effects of tourniquet use in robot-assisted TKA on blood loss, pain, functional recovery, and complications. Methods Patients scheduled for robot-assisted TKA were prospectively randomized into a tourniquet or non-tourniquet group (each n = 14). The primary outcome measure was blood loss. The secondary outcome measures were operation time; visual analog scale (VAS) pain scores; time to achieve the first straight-leg raise; swelling of the thigh, knee, and calf; range of motion; Hospital for Special Surgery score; length of stay; and postoperative complications. Results There was no significant difference in total blood loss between the tourniquet and non-tourniquet groups (738.57 ± 276.158 vs. 866.85 ± 243.422 ml, P = 0.061). The tourniquet group showed significantly lower intraoperative blood loss (P P = 0.002). The non-tourniquet group showed better knee range of motion on postoperative days (PODs) 1–3 (all P P P = 0.044), and shorter length of stay (P = 0.044). Thigh pain VAS score at 1 month after surgery was significantly greater in the tourniquet group (P P P = 0.038). Conclusions Tourniquet use during robot-assisted TKA does not reduce total blood loss, and it appears to increase postoperative pain, aggravate muscle injury, and prolong postoperative recovery. Trial registration ChiCTR, ChiCTR2100041800. Registered 5 January 2021, http://www.chictr.org.cn/index.aspx.