Aim To introduce and explore the potential of digital optical co-site microscopy and image analysis for the observation of changes in dentine surfaces during demineralization. The effect of ethylenediamine tetraacetic acid (EDTA) was evaluated quantitatively and longitudinally. Methodology Three maxillary human molars were sectioned transversely at the cemento-enamel junction, and the crowns discarded. Subsequently, discs approximately 3 mm thick were cut in the cervical third of the root and a standardized smear layer produced. Co-site image sequences of the dentine surface subjected to 17% EDTA were obtained over the experimental period (15, 30, 60, 180 and 300 s). Sixteen images were obtained in each dentine sample for each experimental time, thus, a total of 48 image fields were obtained. For each field, an image analysis routine automatically discriminated open dentine tubules and measured their number, area fraction and minimum diameter, thus allowing the quantification of the demineralization process. The Student t-test was used to analyse the data. Results The number of open tubules remained essentially constant during the demineralization process. The area fraction increased from 9% to 32%. Tubule minimum diameter increased from 1.5 to 3.0 mu m. The changes over time for the area fraction and minimum diameter were significant for comparison between all experimental times (P < 0.05). Conclusions The methodology developed for longitudinal observation of dentinal surfaces was fast, robust and reproducible. It could be easily extended to other chelating substances, thus contributing to the understanding of the demineralization process and in establishing an optimal time-effect relationship in the clinical application.