In 2011 the World Health Organization approved Xpert MTB/RIF for tuberculosis diagnosis and recommended its rapid implementation. Xpert MTB/RIF is accurate: sensitivity is 72.5-98.2% (smear-negative and -positive cases, respectively) and specificity 99.2%. Benefits include same-day diagnosis and simultaneous detection of rifampicin resistance. However, the test has some shortcomings and has not had time for thorough evaluation. Cost-effectiveness studies are difficult to perform and few have been completed. Existing data suggest cost-effectiveness in some, but not all, settings. The urgent need for better diagnostics is evident. Yet, serial implementation of new technologies causes ineffective spending and fragmentation of services. How new tests are incorporated into existing diagnostic algorithms affects both outcomes and costs. More detailed data on performance, effect on patient-important outcomes, and costs when used with adjunct tests are needed for each setting before implementation.While awaiting further clarification it seems prudent to slow its implementation among resource-constrained tuberculosis control programs. Copyright © 2012 by The American Society of Tropical Medicine and Hygiene.