OBJECTIVE To compare the accuracy of clinical examination performed with bitewing radiographs or clinical examination using tooth separation to identify carious lesion activity. METHODS 320 surfaces from 40 bitewing radiographs were examined for approximal caries on the maxillary and mandibular primary molars of 20 patients 3-10 years old. The patients were divided into three groups: (1) Absence of the permanent first molar; (2) Partial eruption of the permanent first molar; and (3) Full eruption of the permanent first molar. Two examiners evaluated the radiographs using a megascope, a magnifying glass (x2), and an amplifying image screen. Approximal radiolucencies were identified on 72 surfaces. Following the radiographic examinations, the two examiners performed conventional clinical inspection using a No. 4 dental mirror, a No. 5 dental explorer, and an air-water syringe, with artificial light and relative isolation. The separation method was performed with elastic bands, which were removed after 24 hours, and the clinical examination conducted as in the non-separation group. RESULTS The correlation between the extension of interproximal radiolucent lesions in primary dentition and their clinical diagnoses following separation of the teeth, was similar to findings on literature evaluating the permanent dentition. On radiographic findings for enamel lesions, white spots predominated both in the inner (100%) and in the outer (94%) half of enamel upon clinical examination with separation of teeth. For radiolucent lesions in dentin, on the other hand, cavities predominated over white spot lesions (84%). In Groups 1 and 2 (young primary), white spots occurred in cases where the radiolucent lesions reached the dentin (15% and 25%), similar to findings for young permanent teeth. Clinical diagnosis performed with the mechanical separation of teeth cannot be considered conclusive for the primary dentition.