Tuberculous peritonitis, although common in Third World countries, remains an uncommon cause of ascites in the United States. Ascitic fluid adenosine deaminase (ADA) activity has been proposed as a useful diagnostic test. The aim of this retrospective study was to determine the clinical utility of ascitic fluid ADA activity in diagnosing tuberculous peritonitis in a U.S. patient population. A total of 368 ascitic fluid specimens from a well-characterized ascitic fluid bank, including tuberculous peritonitis (n = 7), tuberculous peritonitis in the setting of cirrhosis (n = 10), and consecutive specimens of widely varied etiologies (n = 351) were analyzed for ADA activity by ultraviolet spectrophotometry at 265 nm. The overall sensitivity of the ADA determination in diagnosing tuberculous peritonitis was only 58.8%, and the specificity was 95.4%. The accuracy of ADA determination (93.8%) compared favorably with that of the common ascitic fluid tests of white blood cell (WBC) count (>500/mm3), total protein (>2.5 g/dL), and combined WBC count and total protein (45.8%, 74.4%, and 81.3%, respectively). However, ADA was only 30% sensitive in detecting tuberculous peritonitis in the setting of cirrhosis, and cirrhosis was present in 59% of the tuberculous peritonitis patients in our population. In addition, malignancy-related ascites (13%) and bacterial peritonitis specimens (5.8%) occasionally yielded false-positive results. In conclusion, our results indicate that the ascitic fluid ADA activity has good accuracy but poor sensitivity and imperfect specificity in a U.S. patient population in which the prevalence of tuberculosis is low and underlying cirrhosis is common.