It is intuitive that blood loss is lower after unicompartmental knee arthroplasty (UKA), but the difference in potential blood loss between UKA and total knee arthroplasty (TKA) has not yet been studied extensively. The hypothesis of this study was that blood loss is less important in UKA and that it can be performed without transfusion risk, even in preoperative anaemic patients. A comparative matched retrospective study on 105 (57F/48M) UKA patients with a mean (SD) age of 64 (10) years, a mean (SD) BMI of 29.5 (5) kg/m(2) and a mean (SD) preoperative haemoglobin (Hb) level of 13.7 (1.5) g/dl was matched with 105 TKA patients for age, BMI, gender, ASA score and preoperative alignment. Blood loss was measured peroperatively and as a substitute for blood loss by mean Hb and haematocrit drop at day 2 and 4 postoperatively. Mean (SD) visible blood loss was lower for UKA [20 (10) cc vs. 110 (10) cc; P < 0.0001]. Mean (SD) Hb levels were higher for UKA compared to TKA at day 2 [12.9 (1.4) vs. 12.1 (1.4) g/dl; P < 0.0001], day 4 [12.7 (1.4) vs. 11.5 (1.6) g/dl; P < 0.0001] and day 21 [13.2 (1.4) vs. 12.7 (1.3) g/dl; P = 0.0185]. Hidden blood loss was more important for TKA (0.2 vs. 0.6 g/dl). Transfusion rates were 0 % for UKA and 2 % for TKA. Transfused patients were all female with moderate preoperative anaemia (Hb < 11 g/dl) undergoing TKA. Haemoglobin drop is higher with more hidden blood loss in TKA. Patients with preoperative anaemia can undergo UKA without a risk for transfusion; however, women with moderate anaemia planned for TKA are at risk. The transfusion rate for TKA is, however, very low, and so, expected survival of the implant versus risk of co-morbidity should be discussed with the patient. Therapeutic study, Level III.