To determine the sensitivity and specificity of the radiographic detection of calculus, 275 proximal tooth surfaces from 18 patients were evaluated. Standardized periapical radiographs obtained before extraction were coded, batch processed, and evaluated independently by two investigators under optimum viewing conditions. After extraction, the teeth were photographed and evaluated both microscopically and by planimetry on 40 × linear projections. Evaluation of calculus by conventional radiography showed low sensitivity: radiographic deposits were detected on only 44% of surfaces that demonstrated calculus microscopically. Specificity was high and the rate of false positives was only 7.5%. Detection of calculus was influenced by the thickness of calculus, the percentage of root surface occupied by calculus, and by tooth type; but not by attachment loss, probing depth, proximal surface, or arch location. These results show that present radiographic techniques are not appropriate for detecting calculus on root surfaces.