Objective The objective of this work is to compare blood loss during primary knee arthroplasty with the use of intravenous and intraarticular (IV + IA) tranexamic acid versus intraarticular (IA) tranexamic acid alone. Methods This is a randomized, double-blind clinical trial. Patients with indication for primary total knee arthroplasty were recruited in a specialized clinic, where they were operated by the same surgeon, always using the same surgical technique. Thirty patients were allocated in the IV + IA tranexamic acid group and 30 patients in the IA tranexamic acid group, according to randomization. Blood loss was compared through hemoglobin, hematocrit, drain volume, and blood loss estimation (Gross and Nadler calculus). Results After collection, data from 40 patients were analyzed, 22 in the IA group and 18 in the IV + IA group. There were 20 losses due to collection error. Between groups IA and IV + IA, there were no significant differences in 24 hours between hemoglobin levels (10.56 vs. 10.65 g/dL; F 1.39 = 0.63, p = 0.429), erythrocyte (3.63 vs. 3.73 million/mm 3 ; F 1.39 = 0.90, p = 0.346); hematocrit (32.14 vs. 32.60%; F 1.39 = 1.39, p = 0.240); drainage volume (197.0 vs. 173.6 mL; F 1.39 = 3.38 p = 0.069); and estimated blood loss (1,002.5 vs. 980.1; F 1.39 = 0.09, p = 0.770). The same occurred in comparisons conducted after 48 hours postoperatively. Time was a significant factor for the change of all outcome variables. However, the treatment did not modify the effect of time on these outcomes. No individual presented any thromboembolic event during the work period. Conclusions The use of IV + IA tranexamic acid showed no advantage in reducing blood loss when compared to the use of IA tranexamic acid alone in primary knee arthroplasties. This technique proved to be safe, since no thromboembolic event occurred during the development of the work.