The tooth interface with resin-modified glass-ionomer cements (RM GICs) is poorly understood. This study examined the interface, especially with dentin. Cervical cavities in extracted teeth were restored with Fuji II LC, Vitremer, Photac-Fil, or a conventional GIC, Fuji Cap II. Fluorescent dye was placed in the pulp chambers for 3 hrs before the specimens were sectioned. Examination of the tooth/material interface with a confocal microscope showed that dye uptake by the restoration varied among materials. A "structureless", non-particulate, highly-stained layer of GIC was observed next to dentin in Fuji II LC. This layer varied in width, was prominent where the dentin tubules were cut "end-on" and in areas closer to the pulp, and was not seen adjacent to enamel. Vitremer showed minimal dye uptake, and the "structureless" layer was barely discernible. Photac-Fil showed more uniform uptake and absence of this layer. Cracking of enamel was also noted with these materials. The conventional GIC did not show any dye uptake, presence of a "structureless" layer, or enamel cracking. We elucidated the potential mechanisms involved in the formation of a "structureless" interfacial layer in Fuji II LC by studying the variables of cavity design, surface pre-treatment, water content of the tooth, time for it to develop, early finishing, and coating of the restoration. This layer, the "absorption layer", is probably related to water flux within the maturing cement, depending on environmental moisture changes and communication with the pulp in a wet tooth. The "micropermeability model" was useful in this study of the interfacial characteristics of RM GICs.