The PCR technique was applied to the diagnosis of tuberculosis in live cattle, and both skin-test-negative and skin-test-positive animals were studied. DNA was taken from various sources including specimens of lymph node aspirates, milk, and nasal swabs. After slaughter and visual inspection, tissues such as lymph nodes, lungs, and udders from tuberculin reactors were tested by the same technique. Specific oligonucleotide primers internal to the IS6110 insertion element were used to amplify a 580-bp fragment. A 182-bp fragment was obtained by designating a nested PCR from the first amplification product. This fragment was cloned and sequenced, and after being labeled it was employed in dot blot hybridization. A total of 100 cattle were tested, and PCR analysis was performed using nasal swab, milk, and lymph node aspirate. Sixty skin-test-positive cows were also tested to detect mycobacterial DNA in tissue samples from lymph nodes, lungs, and udders, and the infection was confirmed in all of the animals. Using PCR analysis of tissue samples from slaughtered animals as a "gold standard" we calculated 100% values for sensitivity, specificity, and positive and negative predictive values for milk and lymph node aspirate samples. The respective values for nasal swab samples were 58, 100, 100, and 28%. The respective values for all of the samples were 74, 100, 100, and 35%, while for visual inspection the values were 81, 100, 100, and 58%, respectively. PCR analysis of specimens of lymph node aspirates, milk, and nasal swabs from skin-test-negative animals showed that 52% of these skin test results were false negatives. These animals, not being removed from the farms, represent a potential source of further infection.