Plain spine radiographs from 31 patients with pathologically proven intraspinal tumors were analyzed retrospectively by an experienced pediatric radiologist unfamiliar with the cases in order to evaluate the applicability and reliability of the radiographic signs commonly used for diagnosis. Interpediculate distances (IPDs) were graphed onto standard curves to assess whether graphic analysis aided in diagnosis. In this group of patients, no single radiographic criterion permitted better than 40% true-positive diagnosis. False-positive diagnoses were made in 0-10% of cases, depending on the sign employed. Use of multiple roentgen signs in conjunction achieved 55% true-positive and 3% false-positive diagnoses, while use of graphed IPDs alone achieved 57% true-positive and 11% false-positive diagnoses. Addition of graphed IPDs to visual inspection of the radiographs led to detection of 6 tumors (19%) not otherwise appreciated, but tripled the false-positive rate from 3 to 10%.