Currently available nucleic acid amplification platforms for tuberculosis (TB) detection are not designed to be simple or inexpensive enough to implement in decentralized settings in countries with a high burden of disease. The loop-mediated isothermal amplification platform (LAMP) may change this. We conducted a study in adults with symptoms suggestive of TB in India, Uganda, and Peru to establish the feasibility of using TB-LAMP (Eiken Chemical Co.) in microscopy laboratories compared with using smear microscopy against a reference standard of solid and liquid cultures. Operational characteristics were evaluated as well. A total of 1,777 participants met the eligibility criteria and were included for analysis. Overall, TB-LAMP sensitivities among culture-positive samples were 97.2% (243/250; 95% confidence interval [CI], 94.3% to 98.2%) and 62.0% (88/142; 95% CI, 53.5% to 70.0%) for smear-positive and smear-negative TB, respectively, but varied widely by country and operator. Specificities ranged from 94.5% (446/472; 95% CI, 92.0% to 96.4%) to 98.0% (350/357; 95% CI, 96.0% to 99.2%) by country. A root cause analysis identified high temperatures, high humidity, and/or low reaction volumes as possible causes for false-positive results, as they may result in nonspecific amplification. The study was repeated in India with training focused on vulnerable steps and an updated protocol; 580 participants were included for analysis. Specificity in the repeat trial was 96.6% (515/533; 95% CI, 94.7% to 97.9%). To achieve acceptable performance of LAMP at the microscopy center level, significant training and infrastructure requirements are necessary.