This study evaluated the effect of bonded amalgam on the fracture strength of teeth using five adhesive systems: Panavia 21, Amalgambond Plus, Imperva Dual Bond, All-Bond 2 Primer/Bonding Resin, and All-Bond 2 Primer/Liner F. Intact teeth and amalgam lined with Copalite were used as control groups. Large MOD preparations were made in 20 extracted maxillary premolars for each group. The teeth were restored with Tytin. All groups were stored in water at 37 degrees C for 15 days and thermocycled 2500 times, over 8-48 degrees C. The specimens were preloaded five times in compression to 10 kg using a 5 mm-in-diameter, cylindrical steel indenter that contacted the teeth only on the cuspal inclines. Then the teeth were loaded to failure at 5.0 mm/min. The failure mode was recorded (amalgam failure, cusp fracture, or failure at the tooth/amalgam interface). The mean fracture strengths were analyzed using ANOVA and Newman-Keuls multiple comparisons. The Imperva Dual Bond group showed the highest mean forces followed by All-Bond 2 Primer/Bonding Resin. The All-Bond 2 Primer/Liner F and Amalgambond Plus groups demonstrated lower means and were not significantly different from the Copalite group. The Panavia 21 group was in between these two groups and was not statistically different from either group. The mean strength of intact teeth was the highest, but its very large coefficient of variation (60%) prevented effective use of these data for statistical comparison. Analysis of the mode of fracture showed that Panavia 21, All-Bond 2 Primer/Bonding Resin, and Amalgambond Plus failed cohesively in the amalgam in 35%, 25%, and 15% of the specimens respectively. This fracture type is a good indication of effective bonding between tooth and amalgam. The most common type of fracture for all the restored groups was the one that occurred at the tooth/restoration interface. This would suggest that current bonding procedures could be improved.