An ELISA diagnostic test for tuberculosis antigen A60 (TBA60) IgG/IgM was used in a tertiary referral hospital in Taiwan. From June 1992 to December 1993, serum samples obtained from 907 patients were analyzed for TBA60 IgG and IgM titres. The final diagnosis of these patients was confirmed by microbiological study and clinical follow up for 18-24 months. Among 147 patients with active pulmonary tuberculosis, IgG was positive in 112 (76.2%), IgM was positive in 14 (9.52%). Among 90 patients with active extrapulmonary tuberculosis, IgG was positive in 53 (58.9%), IgM was positive in 9 (10%). Among 153 patients with inactive tuberculosis, IgG was positive in 28 (18.3%), IgM was positive in 1 (1.6%). Among 517 patients with nontuberculous disease, IgG was positive in 50 (9.7%), IgM was positive in 3 (0.6%). In this study population with 26% (237/907) active tuberculous infection rate, the TBA60 ELISA IgG had a diagnostic sensitivity of 69.6% and a specificity of 92.1%. These results indicate a positive predictive value of 67.9% and a negative predictive value of 89.2%. The sensitivity of IgM was 10.5% and specificity, 99.4%. The serum IgG tire had good correlation with the extent of pulmonary TB had a higher percentage of IgG seropositivity (83.9%) than those with smear-negative pulmonary TB (70.6%) and extrapulmonary TB (58.9%). In 50 cases with active tuberculosis, follow-up examinations were carried out one month after treatment. In 18 cases with initially negative IgG and IgM titres, 13 showed elevation of serum IgG titres into positive level, one had positive seroconversion of IgM which was the only serological marker indicating active infection. Therefore, 77.8% (14/18) gained diagnostic benefit from follow-up serological examination. It was concluded that TBA60 IgG and IgM ELISA is a useful test when diagnosing tuberculosis. This test also assists in the clinical judgement of tuberculosis when used as an adjunct to symptoms and sputum smear, and for monitoring therapeutic response at the commencement of treatment.