The objective of this study was to measure the efficacy of Mycobacterium smegmatis as a surrogate in vitro model for the detection of compounds which are inhibitory to the growth of Mycobacterium tuberculosis. A chemical screen of the LOPAC library for anti-mycobacterial compounds was performed using M. smegmatis. Parallel screens were conducted with another tuberculosis model, Mycobacterium bovis BCG, and with M. tuberculosis under identical growth conditions and the inhibitors detected across the three species were compared. 50% of compounds that were detected as active against M. tuberculosis were not detected using M. smegmatis compared to 21% of compounds using M. bovis BCG. To examine whether these findings were unique to LOPAC, screens were performed with the NIH Diversity Set and Spectrum Collection. An even higher proportion of M. tuberculosis inhibitors were not detected from the NIH Diversity Set and Spectrum Collection using M. smegmatis compared to M. bovis BCG. These data reveal that a significant proportion of M. tuberculosis inhibitors are missed in library screening with M. smegmatis. The basis of the variation in the inhibitory profiles of M. smegmatis and M. tuberculosis has yet to be fully determined, however, our genomic comparisons indicate that approximately 30% of M. tuberculosis proteins lack conserved orthologues in M. smegmatis compared to 3% being absent in M. bovis BCG. In conclusion, although M. smegmatis offers some technical benefits such as a shorter generation time and negligible risk to laboratory workers, it is significantly less effective in the detection of anti-M. tuberculosis compounds relative to M. bovis BCG. This limitation needs to be taken into consideration when selecting an in vitro screening model for tuberculosis drug discovery.