Early, efficient and inexpensive methods for the detection of pulmonary tuberculosis are urgently needed for effective patient management as well as to interrupt transmission. These methods to detect M. tuberculosis in a timely and affordable way are not yet widely available in resource-limited settings. In a developing-country setting, we prospectively evaluated two methods for culturing and detecting M. tuberculosis in sputum. Sputum samples were cultured in liquid assay (micro broth culture) in microplate wells and growth was detected by microscopic observation, or in Löwenstein–Jensen (LJ) solid media where growth was detected by visual inspection for colonies. Sputum samples were collected from 321 tuberculosis (TB) suspects attending Bugando Medical Centre, in Mwanza, Tanzania, and were cultured in parallel. Pulmonary tuberculosis cases were diagnosed using the American Thoracic Society diagnostic standards. There were a total of 200 (62.3%) pulmonary tuberculosis cases. Liquid assay with microscopic detection detected a significantly higher proportion of cases than LJ solid culture: 89.0% (95% confidence interval [CI], 84.7% to 93.3%) versus 77.0% (95% CI, 71.2% to 82.8%) (p = 0.0007). The median turn around time to diagnose tuberculosis was significantly shorter for micro broth culture than for the LJ solid culture, 9 days (interquartile range [IQR] 7–13), versus 21 days (IQR 14–28) (p<0.0001). The cost for micro broth culture (labor inclusive) in our study was US $4.56 per sample, versus US $11.35 per sample for the LJ solid culture. The liquid assay (micro broth culture) is an early, feasible, and inexpensive method for detection of pulmonary tuberculosis in resource limited settings.