Ninety-six children with an irritable hip possibly representing septic arthritis underwent clinical and laboratory evaluation, radiography, and ultrasonography (US). Forty had normal US scans; none were later shown to have septic arthritis, although three had osteomyelitis. Fifty-six children had effusion at US. Thirty-one of these underwent US-guided aspiration, two unsuccessfully. On the basis of results of aspirate analysis, 15 of these 31 patients underwent surgical drainage, and septic arthritis was diagnosed in all 15. The mean peripheral white blood cell (WBC) count was not significantly higher in patients with septic arthritis (n = 21) than in patients with toxic synovitis (n = 24) (P = .44). However, the erythrocyte sedimentation rate and synovial fluid WBC count were significantly greater in those with septic arthritis than in those with toxic synovitis (P = .002 and P < .0001, respectively). Clinical evaluation, radiographs, and sonograms together determined the need for US-guided aspiration. US-guided aspiration allowed selection of only those with septic arthritis for operative drainage and shortened that procedure. Negative sonograms allow exclusion of septic arthritis but not osteomyelitis.