Between October 1998 and September 1999, 98 patients with symptomatic exudative lymphocytic pleural effusion were enrolled in our study to evaluate the diagnostic sensitivity of polymerase chain reaction (PCR) assay. The mean age was 53.3 years ranging from 18 to 78 years. There were 61 men and 37 women. Pleural fluid was sent for gram staining, AFB staining, aerobic culture, culture of Mycobacterium tuberculosis on LJ media, and cytology. Additional fluid was used for a PCR-assay of the 16 S-23 S rRNA gene spacer sequences and for a nested PCR of the 16 S rRNA gene as a blind control. In cases of free-flow pleural tapping, histopathological analysis was done on three pleural biopsies. Overall etiologies comprised malignancy 53.1%, tuberculosis 36.7%, lymphoma 2.0% and chronic nonspecific inflammation 8.2%. The sensitivity and specificity of AFB-staining were 6% and 79%, respectively; while cultures on LJ media were 17% and 100%, respectively. The sensitivity of the PCR-assay was 50% (95% CI: 40 to 60%) and the specificity was 61% (95 CI: 52 to 71%). When PCR was nested, the sensitivity was 72% (95% CI: 63 to 81%) and specificity was 53% (95% CI: 43 to 63%). Two thirds (26 of 36) of tuberculous pleural effusion cases underwent pleural biopsy, and 62% were diagnosed by histopathology. There were no complications from thoracocentesis or pleural biopsy in any of the patients. We concluded that PCR assay was more sensitive than AFB staining and mycobacteria culture for diagnosis tuberculous pleural effusion but its specificity was quite low.