It is widely acknowledged that new regimens are urgently needed for the treatment of tuberculosis. The primary endpoint in the Phase III trials is a composite outcome of failure at the end of treatment or relapse after stopping treatment. Such trials are usually both long and expensive. Valid surrogate endpoints measured during or at the end of treatment could dramatically reduce both the time and cost of assessing the effectiveness of new regimens. The objective of this study was to evaluate sputum culture results on solid media during treatment as surrogate endpoints for poor outcome. Data were obtained from twelve randomised controlled trials conducted by the British Medical Research Council in the 1970s and 80s in East Africa and East Asia, consisting of 6974 participants and 49 different treatment regimens. The month two culture result was shown to be a poor surrogate in East Africa but a good surrogate in Hong Kong. In contrast, the month three culture was a good surrogate in trials conducted in East Africa but not in Hong Kong. As well as differences in location, ethnicity and probable strain of Mycobacteria tuberculosis, Hong Kong trials more often evaluated regimens with rifampicin throughout and intermittent regimens, and patients in East African trials more often presented with extensive cavitation and were slower to convert to culture negative during treatment. An endpoint that is a summary measure of the longitudinal profile of culture results over time or that is able to detect the presence of M. tuberculosis later in treatment is more likely to be a better endpoint for a phase II trial than a culture result at a single time point and may prove to be an acceptable surrogate. More data are needed before any endpoint can be used as a surrogate in a confirmatory phase III trial.