The purposes of this study were to determine the (1) frequency with which nonenhanced computed tomography (CT) (NECT) permits conclusive diagnosis of acute appendicitis, (2) accuracy of NECT when findings are conclusive, and (3) overall accuracy of a CT protocol consisting of NECT with selective use of contrast. Five hundred and thirty-six patients underwent a NECT protocol with selective use of contrast. Diagnostic accuracy was then determined separately for (1) patients with conclusive initial NECT, (2) patients with inconclusive initial NECT, and (3) all patients. NECT was conclusive on initial interpretation in 404/536 patients and inconclusive in 132/536. Of 132 inconclusive studies, 126 were repeated with contrast (intravenous, oral or rectal). Sensitivity, specificity, and positive and negative predictive value for diagnosis of acute appendicitis were (1) 90%, 96.0%, 84.8%, and 97.4% in patients with conclusive NECT (n = 404); (2) 95.6%, 92.3%, 73%, and 99% in patients with inconclusive NECT followed by repeat CT with contrast; and (3) 91.3%, 95%, 82%, and 98% in all patients. The initial diagnosis of appendicitis may be made by NECT in 75% of patients, with contrast administration reserved for inconclusive NECT studies.