RATIONALE: Development of diagnostic tools with improved predictive value for tuberculosis (TB) is a global research priority. OBJECTIVES: We evaluated whether implementing higher diagnostic thresholds than currently recommended for QuantiFERON Gold-in-Tube (QFT-GIT), T-SPOT.TB and the tuberculin skin test (TST) might improve prediction of incident TB. METHODS: Follow-up of a UK cohort of 9,610 adult TB contacts and recent migrants was extended by re-linkage to national TB surveillance records (median follow-up 4.7 years). Incidence rates and rate ratios, sensitivities, specificities and predictive values for incident TB were calculated according to ordinal strata for quantitative results of QFT-GIT, T-SPOT.TB and TST (with adjustment for prior BCG). MEASUREMENTS AND MAIN RESULTS: For all tests, incidence rates and rate ratios increased with the magnitude of the test result (p<0.0001). Over three years' follow-up, there was a modest increase in positive predictive value (PPV) with the higher thresholds (3.0% for QFT-GIT ≥0.35 IU/mL vs. 3.6% for ≥4.00 IU/mL; 3.4% for T-SPOT.TB ≥5 spots vs. 5.0% for ≥50 spots; and 3.1% for BCG-adjusted TST ≥5mm vs. 4.3% for ≥15mm). As thresholds increased, sensitivity to detect incident TB waned for all tests (61.0% for QFT-GIT ≥0.35 IU/mL vs. 23.2% for ≥4.00 IU/mL; 65.4% for T-SPOT.TB ≥5 spots vs. 27.2% for ≥50 spots; 69.7% for BCG-adjusted TST ≥5mm vs. 28.1% for ≥15mm). CONCLUSIONS: Implementation of higher thresholds for QFT-GIT, T-SPOT.TB and TST modestly increases PPV for incident TB, but markedly reduces sensitivity. Novel biomarkers or validated multivariable risk algorithms are required to improve prediction of incident TB. This article is open access and distributed under the terms of the Creative Commons Attribution Non-Commercial No Derivatives License 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0/).