Setting: There remains a lack of effective and inexpensive diagnostic tools for active tuberculosis (TB) disease. Testing immune responses to proteins secreted by Mycobacterium tuberculosis, such as MPT64, may be a diagnostic option. Objective: To evaluate the sensitivity and specificity of a patch test using MPT64 for the diagnosis of active TB disease. Design: This randomised, double-blind, placebo-controlled, prospective study in Lima, Peru, involved 55 healthy controls and 457 symptomatic individuals referred for routine TB testing by the National TB Control Programme. All subjects underwent a comprehensive diagnostic workup, and received an active patch on one arm and a placebo patch on the opposite arm, which were read after 4 days. Results: Eighty-one (18%) of the symptomatic participants were classified as having definite TB, while an additional 98 (21%) had probable TB. The patch tests performed the same in both groups, with a sensitivity of 27% and specificity of 74%. The area under the receiver operating characteristic curve was 0.495 (95%CI 0.425-0.565). Conclusions: Contrary to existing literature, the MPT64 patch was not sensitive and specific to detect active TB. Given the potential of the test, understanding possible differences in the protein source or underlying genetic factors should be explored further.