Purpose: To evaluate bonding penetration into different enamel substrates., Materials and Methods: Ten extracted human molars were mesiodistally sectioned. Buccal and lingual enamel surfaces were divided into four equal areas using sticky wax. The central two areas of each tooth (n = 20) were demineralized for 12 weeks using an acidic gel (pH 4.8). The lateral areas served as controls. After demineralization, ten specimens were remineralized in a saliva substitute for three weeks. An amine fluoride solution (Elmex Fluid) was applied on one half of each specimen before acid etching. After etching for 120 s, an enamel-bonding agent (Heliobond) containing 0.1% rhodamine was applied onto test and control areas, and was light cured for 60 s. Subsequently, the specimens were sectioned and tag length was determined using a confocal laser scanning microscope (CLSM). Results were statistically compared with ANOVA followed by Scheffe's and Bonferroni/Dunn post hoc tests., Results: With a mean penetration depth of 68 +/- 22 microm, tags in demineralized enamel were significantly longer than in other groups (p < or = 0.01). Penetration decreased significantly in remineralized areas or when fluoride was used (p < or = 0.01), but was still significantly deeper than in control sites (p < or = 0.01)., Conclusion: Penetration of an unfilled resin into enamel was considerably influenced by the degree of dental hard tissue mineralization. Penetration was increased in demineralized enamel; however, remineralized enamel also allowed good penetration of the bonding agent.