OBJECTIVE The aim of this review was to determine the diagnostic accuracy of panoramic radiographic markers in the detection of the relationship between the mandibular canal and third molar roots. METHODS A literature search of electronic databases, Cochrane Oral Health Group's Trials Register, National Research Register, conference proceedings, and abstracts was performed to identify studies that had investigated the diagnostic accuracy of the 3 panoramic radiographic markers (ie, darkening of the root, interruption of the radiopaque borders, and diversion of the mandibular canal). RevMan, version 5.0, and Meta-DiSc software programs were used for the pooled analyses and the construction of a summary receiver operating characteristic curve. RESULTS A total of 5 studies were included, involving 894 observations. The overall pooled sensitivity and specificity for darkening of the root was calculated as 51.2% (95% confidence interval [CI] 42% to 60%) and 89% (95% CI 87% to 90%), respectively. The interruption of radiopaque borders showed a pooled sensitivity of 53.5% (95% CI 78.1% to 81.8%) and a pooled specificity of 80% (95% CI 78.1% to 81.8%). The diversion of the canal criterion had a pooled sensitivity of 29.4% (95% CI 21.8% to 38.1%) and a pooled specificity of 94.7% (95% CI 93.6% to 95.7%). The area under the receiver operating characteristic curve was 70% to 77%. CONCLUSIONS The results of this meta-analysis suggest a reasonable diagnostic accuracy for panoramic radiography in the preoperative evaluation of the relationship between third molars and the canal. Additional studies are needed to examine a more accurate, accessible, and cost-effective initial radiographic technique before third molar surgery.