Background: Sputum from patients with tuberculosis contains subpopulations of metabolically active and inactive Mycobacterium tuberculosis with unknown implications for infectiousness. Methods: We assessed sputum microscopy with fluorescein diacetate (FDA, evaluating M. tuberculosis metabolic activity) for predicting infectiousness. Mycobacterium tuberculosis was quantified in pretreatment sputum of patients with pulmonary tuberculosis using FDA microscopy, culture, and acid-fast microscopy. These 35 patients’ 209 household contacts were followed with prevalence surveys for tuberculosis disease for 6 years. Results: FDA microscopy was positive for a median of 119 (interquartile range [IQR], 47–386) bacteria/µL sputum, which was 5.1% (IQR, 2.4%–11%) the concentration of acid-fast microscopy–positive bacteria (2069 [IQR, 1358–3734] bacteria/μL). Tuberculosis was diagnosed during follow-up in 6.4% (13/209) of contacts. For patients with lower than median concentration of FDA microscopy–positive M. tuberculosis, 10% of their contacts developed tuberculosis. This was significantly more than 2.7% of the contacts of patients with higher than median FDA microscopy results (crude hazard ratio [HR], 3.8; P = .03). This association maintained statistical significance after adjusting for disease severity, chemoprophylaxis, drug resistance, and social determinants (adjusted HR, 3.9; P = .02). Conclusions: Mycobacterium tuberculosis that was FDA microscopy negative was paradoxically associated with greater infectiousness. FDA microscopy–negative bacteria in these pretreatment samples may be a nonstaining, slowly metabolizing phenotype better adapted to airborne transmission.