BACKGROUND:: There is limited data from high-income countries on the performance of interferon-gamma releasing assays in screening for latent tuberculosis infection (LTBI). We analysed the routine application of the Quantiferon-TB Gold (QFT-G) assay, to detect and predict latent and active tuberculosis among HIV-infected patients in Australia. METHODS:: A retrospective cohort study included all HIV-infected patients attending the Melbourne Sexual Health Service between March 2003 and February 2011 who were screened for LTBI using QFT-G. Clinical data was analysed in multivariable models to determine predictors for QFT-G positivity using logistic regression, and active tuberculosis development using Cox proportional hazards. RESULTS:: 917 HIV-infected patients had >/=1 QFT-G performed, of whom 884 (96.4%) were negative, 29 (3.2%) positive, and four (0.4%) indeterminate. The mean age was 40.9 years, 88% were male, with median follow-up of 26.4 (IQR 15.4-30.7) months. 550 (63%) were Australian-born, while 198 (23%) were born in Asia or Africa. QFT-G was positive in 2.0% of Australian-born and 5.3% overseas-born patients (odds ratio[OR] 2.6, 95% CI 1.2-5.6, p=0.017); and 12.7% African-born patients (OR 7.1, 95% CI 2.9-17.3, p<0.001). Two cases of culture-positive tuberculosis occurred after QFT-G screening in 3.4% of QFT-G-positive and 0.1% of QFT-G-negative patients (adjusted hazard ratio 42.4, 95% CI 2.2-827, p=0.013); a rate of 111 (95% CI 27.8-445) per 100,000 person-years:. CONCLUSIONS:: In this context, QFT-G has a high negative predictive (99.9%) value with few indeterminate results. A risk-stratification approach to LTBI screening, where HIV-infected patients with epidemiological risk-factors for tuberculosis infection undergo QFT-G testing, might be clinically appropriate and potentially cost-effective in similar settings.