In low-income and middle-income countries, direct (unconcentrated) sputum smear microscopy is the primary method for diagnosing pulmonary tuberculosis. The method is fast, inexpensive, and specific for Mycobacterium tuberculosis in high incidence areas. The main limitations of direct microscopy are its relatively low sensitivity, especially in individuals co-infected with HIV, and variable quality of the test in programme conditions. Thus, there is a need to identify methods to improve the sensitivity of microscopy. Physical and chemical sputum processing methods, including centrifugation, sedimentation, and bleach, have been studied and found to show promise. We did a systematic review to assess the ability of different processing methods to improve the sensitivity of microscopy. By searching many sources, we identified 83 studies. Overall, by comparison with direct smears, the results suggested that centrifugation with any of several chemical methods (including bleach) is more sensitive, that overnight sedimentation preceded by chemical processing is more sensitive, and that specificity is similar. There were insufficient data to determine the value of sputum processing methods in patients with HIV infection. Operational studies are needed to determine whether the increased sensitivity provided by processing methods is sufficient to offset their increased cost, complexity, and potential biohazards, and to examine their feasibility.