The performance of a colorimetric ADA determination in body fluids other than serum in the diagnosis of tuberculosis was assessed in 1063 patients from whom pleural (600), peritoneal (136), pericardial (77), or cerebrospinal (250) fluids were obtained. In exudative pleuroperitoneal and pericardial effusions, an ADA decision level of 0.71 mu kat/L displayed a sensitivity of 1.00, and was higher than those of histologic (0.83) and bacteriologic (0.62) studies. At this level, ADA reached a specificity of 0.92 and efficiency of 0.94. In cerebrospinal fluid, an ADA catalytic concentration above 0.15 mu kat/L strongly suggests tuberculous meningitis in patients older than 7 years (sensitivity 1.00, specificity 0.99 and efficiency 0.99). ADA results obtained with a UV-method were closely correlated with those of the colorimetric method in pleuroperitoneal effusions (r = 0.989) and in cerebrospinal fluids (r = 0.905). Sample blanks should be processed, otherwise false positive results may be found in non-tuberculous cerebrospinal fluids (5.3%) and pleuroperitoneal effusions (3.8%).