AIMS: We assessed and compared the diagnostic accuracy of two sets of diagnostic criteria for headache secondary to temporomandibular disorders (TMD). METHODS: In 373 headache subjects with TMD, a TMD headache reference standard was defined as: self-reported temple headache, consensus diagnosis of painful TMD and replication of the temple headache using TMD-based provocation tests. Revised diagnostic criteria for Headache attributed to TMD were selected using the RPART (recursive partitioning and regression trees) procedure, and refined in half of the data set. Using the remaining half of the data, the diagnostic accuracy of the revised criteria was compared to that of the International Headache Society's International Classification of Headache Diseases (ICHD)-II criteria A to C for Headache or facial pain attributed to temporomandibular joint (TMJ) disorder. RESULTS: Relative to the TMD headache reference standard, ICHD-II criteria showed sensitivity of 84% and specificity of 33%. The revised criteria for Headache attributed to TMD had sensitivity of 89% with improved specificity of 87% (p < 0.001). These criteria are (1) temple area headache that is changed with jaw movement, function or parafunction and (2) provocation of that headache by temporalis muscle palpation or jaw movement. CONCLUSION: Having significantly better specificity than the ICHD-II criteria A to C, the revised criteria are recommended to diagnose headache secondary to TMD.