OBJECTIVE To evaluate the potential of recombinant 38000 protein of Mycobacterium tuberculosis (38000 protein) as a tuberculosis-specific tuberculin for screening M. tuberculosis infection. METHODS A total of 1342 subjects (706 men and 636 women, age 18-60 years) from several communities in Kazuo County and Xidaziying Town, Chaoyang, Liaoning Province, and Hongdong County, Linfen, Shanxi Province were enrolled from September 2004 to February 2005. The skin tests were performed with double-blinded and the intradermal injections were administered on both forearms with 0.1 ml solution of PPD and 38000 protein at the right side and the left side, respectively. The vertical and transverse diameters of induration or erythema were measured following 24 h for 38000 protein and 48 h for PPD, respectively. The diameters of the the skin test reactions were defined as the means of the vertical and transverse diameters, and positive skin reactions were identified when the diameter was greater than or equal to 5 mm. The comparison of the positive rate was performed via chi(2) test and the consistency of positive skin test reactions between 38000 protein and TB-PPD was analyzed through calculating Kappa coefficients. RESULTS The positive rate was 55.1% (740/1342) and 28.6% (384/1342) for PPD and 38000 protein, respectively; the difference being significant (chi(2) = 190.6, P < 0.01). The consistency of positive skin test reactions between 38000 protein and PPD was low due to a negative Kappa coefficient. The positive rates induced by PPD and 38000 protein tended to increase with age except for the 33-37 year group. For a given age group, the positive rate of PPD was much higher than that of 38000 protein. The subjects without BCG scar had a lower positive rate for 38000 protein (24.3%, 137/566) than those with BCG scar (31.9%, 247/776) (chi(2) = 4.7, P < 0.05). The subjects with tuberculosis contact history had a higher positive rate for 38000 protein (74.4%, 32/43) than those without tuberculosis contact history (27.1%, 352/1299). Subjects without tuberculosis contact history had a lower positive rate for 38000 protein (27.1%, 352/1299) than that of PPD (54.0%, 702/1299), all of which showed significant difference (chi(2) test values changed from 4.7 to 192.1, P < 0.05 or P < 0.01). For those with a tuberculosis contact history, no significant difference in the positive rate was found between 38000 protein (74.4%, 32/43) and PPD (88.4%, 38/43) (chi(2) = 1.9, P > 0.05). Nor was significant difference found in the positive rate of 38000 protein between male subjects (28.5%, 201/706) and female subjects (28.8%, 183/636). The diameter of the positive reactions induced by 38000 protein and PPD ranged from 5 - 9 mm and from 5 - 14 mm, respectively. CONCLUSIONS Our findings show that 38000 protein may be useful as a tuberculosis-specific skin test antigen for screening Mycobacterium tuberculosis infection.