To examine the reduction in enamel demineralization provided by fluoride release from a conventional glass-ionomer, a resin-modified glass-ionomer and an experimental fluoride-releasing resin-based composite compared to a conventional resin-based composite control, and to correlate the level of fluoride release with demineralization. Enamel surfaces of extracted human incisors had a 0.4 mm thick layer of the specified test material carefully placed in a band across the mid-facial enamel to simulate a cement layer beneath an orthodontic bracket. The top surface of the test material was covered with nail varnish, leaving only the edges of the material exposed to release fluoride. The teeth were additionally covered with nail varnish to within 1 mm of the test material. Each group of teeth was placed into separate volumes of unstirred demineralizing solution at a pH of 4.7 for 4 days. The specimens were sectioned and examined by polarized light microscopy. Lesion areas were measured at distances from 100-800 microm away from the test material. Fluoride release for the test materials was measured for periods up to 5 months. All of the fluoride-releasing materials demonstrated a statistically significant (P< 0.05) degree of protection of enamel from demineralization compared to the non-fluoride control material. The degree of protection was greatest near the material, but lesion areas increased with distance in an inverse relationship to the amount of fluoride release. Lesions were displaced from the region near the materials and the mean displacement was directly related to amount of fluoride release. The mean lesion areas for each distance decreased with the logarithm of the cumulative fluoride release.