Patients with moderate or severe pain following knee arthroplasty and washout from standard patient-controlled analgesia (PCA) were randomized to receive 20 mg of an extended-release (ER) oxymorphone formulation (n = 65) or placebo (n = 61) q12h for 1 day. Oxymorphone PCA was used as rescue analgesic. Oxymorphone ER provided significant improvements over placebo for most standard single-dose analgesic parameters, including mean total pain relief (TOTPAR) over 0 to 12 hours (19.30 vs. 13.72; p = 0.0056), as well as for all multiple-dose (24-h) efficacy assessments. Oxymorphone-treated patients used significantly less rescue PCA than those who received placebo (p < 0.02). Adverse events such as nausea and constipation were typical of opioids, and laboratory and physical findings were similar between groups. Oxymorphone ER was effective and generally well tolerated. A single dose was active from 2 hours until > or = 12 hours after administration. Comparisons with other oral opioids are warranted, especially in the setting of outpatient and day surgery.