The effect of quantum noise on the radiographyic detectability of proximal carious lesions was studied. Bitewing radiographs obtained from patients considered to be free of caries were digitized and stored in a computer using a sampling interval comparable to screen-film systems. Lesions limited in depth to one half the thickness of the enamel were induced by means of computer simulation. Various amounts of randomly generated quantum noise were added to each radiographic image. The results from observer-performance studies show that the lesions can be detected reliably in the presence of substantial amounts of quantum noise, suggesting that comparable diagnostic performance could be obtained with significantly faster imagining systems than those currently used in dentistry.