Purpose: The purpose of this study was to compare approximal carious lesions in dentin diagnosed by clinical examination with those diagnosed radiographically in high caries-risk children. Methods: Eighty-four Amish two- to 11-year-olds were evaluated clinically by calibrated examiners using standard epidemiological criteria and radiographically using standard criteria as part of baseline examinations for a two-year pilot clinical trial. Approximal carious surfaces diagnosed by clinical examination were compared to those diagnosed radiographically to determine the number of lesions misdiagnosed by clinical examination alone. Comparisons between the number of lesions detected by clinical exam and radiographic exam were made using McNemar's test. Results: Among 124 anterior approximal tooth surfaces, clinical examination alone identified 61 lesions (49.2 percent), while radiograph examination identified 72 lesions (58.1 percent); thus, nine percent of lesions were misdiagnosed by clinical examination alone. For the 412 posterior approximal tooth surfaces included in the study, 101 lesions (24.5 percent) were identified by clinical examination, whereas 229 lesions (55.6 percent) were identified via radiographic examination. Thus, 128 (31.1 percent) were misdiagnosed by clinical examination alone. Conclusions: Clinical examinations without radiographs significantly underestimate caries prevalence. Therefore, the use of radiographs should be encouraged in epidemiological studies and clinical trials.