In spite of the excellent performance of rapid tuberculosis (TB) nucleic acid amplification (NAA) tests and the clear benefits of immediate diagnosis of TB disease, NAA tests frequently are not used in the diagnosis of pulmonary TB cases, particularly TB cases with smear-negative sputa. Public health laboratories primarily perform TB NAA tests only on a targeted subset of specimens, usually including those that are smear positive and those for which a clinician has specifically requested NAA testing. As an alternative to targeted testing, some laboratories use TB NAA tests universally for all respiratory specimens, though this practice can be prohibitively costly and can be associated with an increased frequency of false-positive results due to testing of lower-risk patients. We propose a strategy for identifying individuals for NAA testing on the basis of nonclinical risk criteria that are routinely provided on the test requisition form, such as type of health care facility from which the specimen is received and patient age group. Use of this strategy at the Massachusetts Department of Public Health Laboratory would allow for NAA test identification of approximately 54 (74%) of 72 culture-positive pulmonary TB cases over a 1-year period while requiring NAA testing for only 933 (17%) of 5,469 individuals submitting respiratory specimens. We demonstrate that use of nonclinical NAA test selection criteria is an effective strategy for maximizing the number of TB cases that can be rapidly identified while minimizing the number of specimens that must be tested.