Controversy exists as to whether there is less secondary caries at the margins of glass ionomer restorations compared with other materials that do not release fluoride. This study examined the incidence of secondary caries for three types of restorative materials in Class V restorations in xerostomic patients. The study group consisted of 45 high caries-risk adult patients who had undergone head and neck irradiation for the treatment of cancer. All were substantially xerostomic and in need of at least three restorations in the same arch. Every patient received a restoration with each of the test materials, a conventional glass ionomer (GI), a resin modified glass ionomer (RMGI) and a resin composite (C). Patients were instructed in the daily use of a neutral pH sodium fluoride gel in custom trays. Recall appointments were made at 6, 12, 18 and 24 months, and the restorations were examined for material loss, marginal integrity and recurrent caries at the restoration margin. Fluoride compliance was determined at each recall period and recorded as the percentage of recommended use during that interval. Patients were categorized at the end of the study as fluoride non-users if their average compliance was 50% or less. Those with greater than 50% compliance were categorized as fluoride users. In the latter group, no recurrent caries was found for any of the restorations, whereas a material-dependent incidence of recurrent caries was found in the fluoride non-user group. None of the GI, one RMGI and eight C restorations failed due to recurrent caries. For the fluoride non-user patients, Fishers exact test (p=0.05) showed no statistical difference between GI and RMGI but statistical differences were found among those materials and resin composite at each recall period. Recurrent caries reductions for GI and RMGI relative to C were greater than 80% in xerostomic patients not using topical fluoride supplementation.