Mouse and guinea pig models have been used to identify Mycobacterium tuberculosis mutants attenuated for survival. However, unlike mice, M. tuberculosis-infected guinea pigs form caseating granulomas, which may simulate human disease more closely. We used designer arrays for defined mutant analysis, a high-throughput subtractive competition assay, for genotypically defined M. tuberculosis mutants and compared the survival of the same mutant pools in guinea pig and mouse aerosol models. Selected mutants found to be attenuated in either aerosol model were also analyzed in the mouse hollow-fiber model. M. tuberculosis mutants representing 74 genes were tested. Eighteen M. tuberculosis mutants were attenuated for survival in either aerosol model, with 70% of selected mutants also attenuated in the mouse hollow-fiber model. The majority of attenuated mutants in the mouse aerosol model were detected only after 90 days of infection. There was a high degree of concordance between the genes identified by the 2 aerosol models, with detection being significantly earlier in the guinea pig (P<.0003). We identified M. tuberculosis genes required for survival in mammalian lungs. The majority of mouse late-stage survival mutants were detected significantly earlier in the guinea pig, which suggests that differences in tuberculosis-induced lung pathologic changes may account for this accelerated detection.