In a large multicenter study involving six major study sites in the United States, Canada, and Europe, the susceptibilities of 272Mycobacterium tuberculosis strains to classical second-line antituberculosis (anti-TB) drugs (capreomycin, cycloserine, ethionamide, and kanamycin) and newer compounds (amikacin, clofazimine, ofloxacin, and rifabutin) were determined by the radiometric BACTEC 460 procedure and the conventional proportion method on Middlebrook 7H10 agar. Previously established critical concentrations for classical second-line anti-TB drugs were compared with several concentrations in liquid medium to establish equivalence. MICs of newer compounds determined in liquid medium were either the same or up to four times lower than those determined in agar medium. After establishing critical concentrations (breakpoints) in the extended testing of clinical isolates, we obtained an excellent overall correlation between the two systems, with no errors with amikacin, kanamycin, and ofloxacin and very few major or very major errors with the other drugs; however, for cycloserine, no breakpoint concentration could be recommended due to repeatedly inconsistent results by both methods. Based on these data we conclude that the BACTEC 460 procedure is a simple and rapid method requiring 4 to 8 days on average to generate accurate antimicrobial susceptibility testing (AST) results for eight anti-TB drugs other than those considered primary ones. These data not only fill a major gap of knowledge regarding the critical test concentrations of secondary anti-TB drugs but also provide a baseline for future evaluations ofM. tuberculosis AST with the more recently developed, nonradiometric broth-based culture systems.