Forty-five patients were evaluated during knee arthroscopy performed using local anesthesia produced by lidocaine with epinephrine to determine the dose-response relationship for operative analgesia. Serum lidocaine concentrations were also measured. Patients were randomized prospectively to receive 20 mL of 0.5%, 1.0%, or 1.5% lidocaine with epinephrine intraarticularly. Intraoperative discomfort was measured by verbal response on an 11-point linear pain scale. Pain scores were significantly higher in patients receiving 0.5% lidocaine during the first 45 min of surgery (P = 0.03). After 45 min, pain scores continued to be higher in the 0.5% lidocaine group than in the 1.0% or 1.5% groups, but the differences were not statistically significant. Ninety-four percent of patients in the 1.5% lidocaine group were willing to repeat this anesthetic technique for surgery compared with 83% of those in the 1.0% lidocaine group and 75% of those in the 0.5% lidocaine group (P greater than 0.05). The duration of postoperative analgesia was similar in all groups. Serum lidocaine concentrations before and 15, 30, 60, and 120 min after instillation of lidocaine were highest in the 1.5% lidocaine group with a peak concentration of 278 ng/mL. No patient had symptoms of lidocaine toxicity. We recommend that lidocaine concentrations of 1.0% or 1.5% be used when 20 mL is instilled intraarticularly for knee arthroscopy based on patient comfort and absence of lidocaine toxicity.