BACKGROUND: In South Asia, it is estimated that 80% of patients choose to attend a private facility for their healthcare needs. Although patients generally believe that the private-sector provides high quality services, private diagnostic laboratories are largely unregulated and little is known about the accuracy of results provided. This study assesses the accuracy of sputum smear microscopy for pulmonary tuberculosis diagnosis in private laboratories operating in Karachi, Pakistan. A novel evaluation methodology was designed in which patient-actors submitted sputum specimens spiked with cultured Mycobacterium tuberculosis (Mtb) for testing such that laboratories were not aware that they were being assessed. METHODS: Smear-negative sputum specimens from Indus Hospital TB Program patients were collected and combined with an attenuated, cultured Mtb strain to create Mtb-spiked samples; for negative standards, no Mtb was added to the smear-negative sputum specimens. Seven of the largest private laboratories across Karachi were chosen for evaluation and were sent six Mtb-spiked and one Mtb-negative sputum specimens. Patient-actors pretending to be laboratory customers submitted these specimens to each laboratory for testing over a three day period. RESULTS: Only three laboratories accurately classified all the Mtb-spiked specimens which were submitted. A further three misclassified all the Mtb-spiked specimens as smear-negative, thus providing the 'patients' with false negative results. CONCLUSIONS: TB sputum smear microscopy services are highly variable across private laboratories and are often of extremely poor quality. Engagement, capacity building and rigorous monitoring of standards at private laboratories are of vital importance for the control of TB. Our findings, while specific for TB diagnostic tests, could be symptomatic of other tests performed in private laboratories and warrant further investigation.