Objective: The purpose of this study was to test the support of occlusal enamel by amalgam (Dispersalloy), bonded amalgam (AmalgamBond Plus adhesive system with high-performance additive powder and Dispersalloy amalgam), and a bonded resin composite (Scotchbond Multi-Purpose adhesive system and Z100 resin composite). Method and materials: For each material, 16 human molars were prepared. The lingual cusps were removed, and the dentin supporting the facial cusps was cut away, leaving a shell of enamel. Each group of prepared teeth was restored using the materials according to manufacturers' instructions. All specimens were thermocycled (1,500 cycles, 6°C to 60°C, dwell time 30 seconds), then mounted vertically in cast stone. A separating disk, mounted horizontally, was used to create a flat horizontal surface approximately 1.5 mm wide at the tips of both remaining enamel cusps of each specimen. Specimens were loaded evenly across the cusp tips at a crosshead speed of 10 mm/min until the fracture occurred. Results: Data were analyzed using analysis of variance and Tukey A post hoc analysis. No significant difference was detected among amalgam, bonded amalgam, and bonded resin composite in their ability to support occlusal enamel. Support provided for occlusal enamel by the materials, whether bonded or not, was intermediate between no support and that provided by dentin. Conclusion: Without further development of materials technology and evidence of efficacy, restorative materials should not be relied upon to support undermined occlusal enamel.