Abstract
A total of 22 extracted human molars and premolars with a total of 34 initial proximal carious lesions were radiographed in 16 different views varying the horizontal angulation of the X-ray beam in steps of 2.5.degree.. The radiographs were read simultaneously by 3 observers. Clinical and radiographic diagnoses were compared. When utilizing all 16 views the concordance was 95%. All proximal surfaces were scored using 4 scoring classes. Surfaces without radiolucencies were scored 0 and carious surfaces were scored according to the extent of the radiolucencies. Nine sound and 12 carious surfaces were assigned identical scores in all 16 views, while 22 (65%) of the carious surfaces were assigned 2, 3 or 4 different scores. Deviations from a direction of the X-ray beam tangential to the proximal surface eliciting a radiographic image belonging to a different score were measured. In 71% of the cases a deviation of 7.5.degree. or less elicited a different score. Projectional circumstances should be taken into consideration in interpreting proximal radiolucencies.