The purpose of the present investigation was to compare the accuracy of treatment decisions in proximal sites using three intra-oral radiographic systems. Additionally, the impact of an automated non-linear grey-level display was evaluated. Ten observers assessed 84 surfaces on bitewing radiographs for their requirement of restorative treatment using a 6-rank confidence scale. Radiographs were taken with conventional film images (Ultraspeed), a storage phosphor plate (Digora) and a CCD system (Dexis). Additionally, the Dexis software was expanded by a contrast enhancement routine (Dexis+). The restorative treatment threshold was defined as presence of macroscopic cavitation. Regarding the areas below mean ROC curves no significant differences were detected between the groups (p>0.05). Likelihood ratios for positive test results were: 5.29 (Ultraspeed), 8.14 (Digora), 9.67 (Dexis) and 11.37 (Dexis+). The accuracy of restorative treatment decisions based on digital and conventional radiographs is comparable. If a dichotomous treatment decision was requested, the digital systems demonstrated a notable tendency towards higher likelihood of true-positive decisions.