Forty sound upper molars were allocated randomly into four groups of 10 teeth. After flattening occlusal enamel, undermined cavities were prepared in thirty teeth to make three types of specimens with various thickness of occlusal structure (Group 1 ~ 3). All the cavity have the 5 mm width mesiodistally and 7 mm depth bucco-lingually. Another natural 10 teeth (Group 4) were used as a control group. Teeth in group 1 have remaining occlusal structure about 1 mm thickness, which was composed of mainly enamel and small amount of dentin. In Group 2, remained thickness was about 1.5 mm, including 0.5 mm thickness dentin. In Group 3, thickness was about 2.0 mm, including 1 mm thickness dentin. Every effort was made to keep the remaining dentin thickness about 0.5 mm from the pulp space in cavitated groups. All the thickness was evaluated with radiographic Length Analyzer program.

After acid etching with 37% phosphoric acid, one-bottle adhesive (Single Bond™, 3M/ESPE, USA) was applied following the manufacturer's recommendation and cavities were incrementally filled with hybrid composite resin (Filtek Z-250™, 3M/ESPE, USA). Teeth were stored in distilled water for one day at room temperature, after then, they were finished and polished with Sof-Lex system.

All specimens were embedded in acrylic resin and static load was applied to the specimens with a 3 mm diameter stainless steel rod in an Universal testing machine and cross-head speed was 1 mm/min. Maximum load in case of fracture was recorded for each specimen.

The data were statistically analyzed using one-way analysis of variance (ANOVA) and a Tukey test at the 95% confidence level.

The results were as follows:

Within the limits of this study, it can be concluded the fracture resistance of the undermined cavity filled with composite resin was lower than that of natural teeth, however remaining tooth structure may be supported and saved by the reinforcement with adhesive restoration, even if that portion consists of mainly enamel and a little dentin structure.