We compared three different methods of anesthesia for outpatient knee arthroscopy in terms of perioperative surgical conditions, pain, and hemodynamics. In a prospective and double-blind study (n=130) the patients were randomized into three groups. A 50-ml mixture composed of 20 nil 0.5% bupivacaine hydrochloride, 10 ml 2% lidocaine hydrochloride, and 20 nil 0.9% sodium chloride was prepared for local anesthesia. The knee joint was injected with 40 nil of the mixture. The portal sites were then injected with 10 ml of the mixture in group I. Using the same technique 250 mug epinephrine was added to the same mixture in group II. In group III the knee joint was injected with 40 nil of the mixture, and only 50 mug epinephrine was then added to 10 ml of the mixture left before the portal site injections, A tourniquet was not used. There were some statistically significant changes in hemodynamic data. Also the data on visual analogue scale scores, time of arthroscopy, and amount of liquid used for intra-articular flushing in group 11 and III were significantly lower than those in group I. According to our experience. bleeding in arthroscopy comes mostly from portal incision to intra-articular field, except when performing extensive synovial shaving, ligament reconstruction, and lateral retinacular release. Therefore, when hemostasis is obtained at portals, the arthroscopic view becomes clearer. We think that adding epinephrine to only portal site injections is sufficient to obtain a clear view and, furthermore, when carrying out arthroscopy in this manner, no significant changes are encountered in heart rate, mean arterial pressure, pain during arthroscopy, or time of arthroscopy.