A total of 1208 positive BACTEC vials were examined for the presence or absence of serpentine cording. A very high (92.9%) rate of laboratory prevalence was obtained for Mycobacterium tuberculosis complex. The sensitivity, specificity, positive and negative predictive values of this test were 92.7%, 95.3%, 99.6% and 50.0%, respectively. It was concluded that testing cord formation in laboratories that have a high prevalence of Mycobacterium tuberculosis complex is an exceptionally reliable method for preliminary reporting of cording-positive cases; however, for cording-negative cases, preliminary reports based solely on cord formation are not reliable. It was also observed that the length of the incubation period has a significant effect on cord formation. Incubation periods of 4 days or less are not sufficient to determine noncording in smears prepared from positive BACTEC vials.