To compare the effects of different image processing modes (filters) on the reproducibility and validity of the assessment of interproximal carious lesions viewed in digitized radiographic images.Variables studied were film type, exposure time, and tissue scatter radiation. Images were assessed in relation to film type, exposure time and tissue scatter. Standardized radiographs of 50 extracted teeth were obtained and 44 areas of interproximal caries were detected. Six radiographs were obtained of each tooth and digitized using FRIACOM software. Four were made using D-type film with and without soft tissue scatter equivalent at normal exposure time (0.32 seconds) and underexposed (0.16 seconds). Two were made using E-type film with TSE normal (0.16 seconds) and underexposed (0.08 seconds). On each of the 300 radiographs, the central depth (CD) of each carious lesion was measured at 18x magnification both on the unchanged radiographic image and after use of two different filters. All measurements were repeated after 1 week to estimate reproducibility.Histometric CD assessments provided a gold standard for comparison with the radiographic measurements (validity). Whereas reproducibility was not influenced by one of the filters, filter "structure" improved the overall validity of CD measurements. In this study, digital manipulations of radiographic images resulted in statistically significant improvement of the validity of CD measurements.