Most of the world's tuberculosis cases occur in low-income and middle-income countries, where sputum microscopy with a conventional light microscope is the primary method for diagnosing pulmonary tuberculosis. A major shortcoming of conventional microscopy is its relatively low sensitivity compared with culture, especially in patients co-infected with HIV. In high-income countries, fluorescence microscopy rather than conventional microscopy is the standard diagnostic method. Fluorescence microscopy is credited with increased sensitivity and lower work effort, but there is concern that specificity may be lower. We did a systematic review to summarise the accuracy of fluorescence microscopy compared with conventional microscopy. By searching many databases and contacting experts, we identified 45 relevant studies. Sensitivity, specificity, and incremental yield were the outcomes of interest. The results suggest that, overall, fluorescence microscopy is more sensitive than conventional microscopy, and has similar specificity. There is insufficient evidence to determine the value of fluorescence microscopy in HIV-infected individuals. The results of this review provide a point of reference, quantifying the potential benefit of fluorescence microscopy, with which the increased cost and technical complexity of the method can be compared to determine the possible value of the method under programme conditions.