OBJECTIVE To assess whether inflammation on ultrasound is predictive of clinical response to intraarticular (IA) corticosteroid injections in patients with knee osteoarthritis (OA). METHODS Patients with symptomatic knee OA were randomized to receive either an IA injection of 40 mg triamcinolone acetonide in the treatment group or 1 cc 0.9% saline in the placebo group. Clinical response was assessed by changes in baseline Western Ontario and McMaster Universities (WOMAC) index scores and physician global assessment at 4 and 12 weeks. Ultrasounds were performed at each visit. Patients and assessors were blinded to treatment status. RESULTS Seventy-nine patients were enrolled into the study. Four-week data were available for 67 patients in the primary analysis comparing change in WOMAC pain score from baseline to 4 weeks. There was almost no change in the WOMAC pain subscale score from baseline to 4 weeks in the control group, but there was a significant improvement in WOMAC pain subscale score from 10.8 (SD +/- 3.2) at baseline to 8.75 (SD +/- 4.0) at 4 weeks in the treatment group (adjusted p = 0.001). Of the 34 patients in the treatment group; 16 (47%) had inflammatory disease and 18 (53%) had noninflammatory disease as determined by ultrasound. There was no difference in the change in WOMAC pain score between the inflammatory and noninflammatory patients in the treatment group at 4 weeks. There was a statistically significant greater improvement in pain subscale scores among noninflammatory patients than among inflammatory patients at 12 weeks. CONCLUSIONS Intraarticular corticosteroid injections are an effective short-term treatment for symptomatic knee OA compared to placebo. Patients with noninflammatory characteristics on ultrasound had a more prolonged benefit from IA corticosteroids compared to inflammatory patients.