BACKGROUND To guard against the potential risk of cusp fracture, esthetic onlay restorations have been advocated for teeth with large restorations. The influence of the adhesive resin cement is believed to play a role in strengthening these restorations. The durability of this tooth/adhesive/ceramic interface is critical to ensure clinical longevity. OBJECTIVE The purpose of this study was to assess the effects of cyclic loading and environmental aging on the residual interface strength of a ceramic bonded to dentin structure. METHODS Eighteen simple trilayer specimens were fabricated, consisting of a 1.5-mm-thick ceramic plate (ProCAD) bonded to a flattened human molar tooth with exposed coronal dentin. The ceramic plates were bonded using resin cement (Nexus 2) and manufacturer-recommended bonding techniques. The specimens were divided into 3 equal groups and were stored in water at 37 degrees C for 10 weeks as a control group (CT), 9 months as an aging group (AG), or placed in water at 37 degrees C while being subjected to 10 million vertical loading cycles between 20 N to 200 N, as a fatigue group (FG). After the specimens were subjected to the experimental conditions, they were sectioned perpendicular to the flat ceramic surface into 1 x 1-mm sticks. The mean residual interface microtensile bond (MTB) strength was determined for each specimen using only those sticks which contained ceramic bonded to dentin. The MTB strength data were analyzed using Weibull analysis methods to determine differences between groups. All subsequent failed specimen surfaces were evaluated under a stereomicroscope at x10 magnification to determine the apparent failure modes. Some specimens were selected from each failure mode category for surface evaluation under a scanning electron microscope (SEM). RESULTS The characteristic Weibull means for the 3 groups were CT, 19.2, FG, 14.7, and AG, 11.7. The bond strength of group CT was significantly greater than both AG (P=.007) and FG (P=.014). Light microscopic categorization of the failure modes suggests that adhesive failure at the ceramic/cement interface was the most common (65%) for all 3 groups. SEM evaluation of failed surfaces of select specimens from each group could not distinguish any interface appearance differences. CONCLUSIONS For indirect adhesive-retained ceramic restorations, both cyclic masticatory loading and hydrolytic degradation may contribute to a weakening of the interface bond. The ceramic/resin interface may be more susceptible to these changes over the time frame of this investigation than the dentin/resin interface.