During 13 years 1288 patients were tested with commercial tuberculin (PPD-T) and nontuberculous mycobacterial (NTM) purified protein derivative (PPD) antigens. Five hundred eighty-three (45%) had findings to suggest tuberculous infection or were asymptomatic tuberculin (PPD and/or old tuberculin) test reactors; the remaining 705 (55%) were PPD nonreactors. On the initial battery of PPD tests 553 patients had positive reactions; 513 (94%) could be classified as a human tuberculous reactor or a nontuberculous reactor. Of the 553 patients 363 were asymptomatic and 190 had clinical disease. Fifty-nine patients had Mycobacterium tuberculosis (TB) disease; 131 had NTM disease. Mycobacterial isolates recovered in 71 (47%) of 150 patients cultured were compared to the initial and repeat PPD-T and NTM-PPD test results. All 22 patients with TB isolates were positive to both PPD-T and at least one NTM-PPD antigen, while 36 of 42 patients (86%) had a positive homologous NTM-PPD reaction compared to that of the NTM species isolated. A PPD-T reaction of 5 to less than 15 mm suggested either a TB or NTM infection, while a PPD-T greater than or equal to 15 mm was strongly associated with TB infection. Dual (PPD-T and PPD-Battey) Mantoux testing in tuberculin-positive children and adolescents can discriminate (88% of 26 Mycobacterium avium-intracellulare culture-positive subjects) between TB or NTM infections and should be a valuable guide in their management.