In this study, we evaluated the diagnostic accuracy and cost-effectiveness of ascitic fluid interferon-γ (IFN-γ) and adenosine deaminase (ADA) assays in the diagnosis of tuberculous ascites. Ascitic fluid from patients with proven tuberculosis (TB) (n = 31) and non-TB ascites (n = 88) was analyzed for IFN-γ and ADA levels. Areas under the receiver operative characteristic (ROC) curves (AUCs) for the two biologic markers were compared. Levels of ascitic fluid IFN-γ, median (range): 560 (104-1600) pg/mL vs. 4.85 (0-320) pg/mL (p < 0.001), and ADA, median (range): 58 (16-331) IU/L vs. 10 (0-59) IU/L (p < 0.001), were significantly different between TB and non-TB groups. IFN-γ and ADA assays showed equal sensitivity (0.97) and differed marginally in specificity (0.97 vs. 0.94). Difference in AUCs was not significant (0.99 vs. 0.98, p = 0.62). For differentiating TB from non-TB ascites, optimal cutoff points were 112 pg/mL for IFN-γ and 37 IU/L for ADA. The accuracy of the ADA assay was similar to that of the IFN-γ assay in differentiating of TB from non-TB ascites. Because both material and human costs of the ADA assay are far less than those of the IFN-γ assay, the former is probably the most appropriate diagnostic test for analysis of peritoneal fluid in resource-limited settings.