Objective: This study evaluated the load required to produce failure in Class II bonded amalgam restorations. Method and materials: Five groups of 12 maxillary molars were mounted. Class II mesio-occlusal preparations were cut: group 1, extension through central grooves, wihtout retention grooves; group 2, proximal slot preparation, without retention grooves; groups 3 and 5, slot preparation with long facial and lingual retention grooves; group 4, slot preparation without grooves, with unsupported proximal enamel allowed to remain. Groups 1 to 4 were restored with an amalgam-bonding system and amalgam; in group 5, no amalgam-bonding material was used. Results: Mean (SD) failure loads: group 1, 281 (77) N; group 2 , 246 (101) N; group 3, 238 (84) N; group 4, 254 (100) N; and group 5, 191 (66) N. Conclusion: Although there was a trend toward greater resistance to dislodgement or fracture when an amalgam-bonding technique was used, its use did not significantly increase the resistance to failure when compared to retention grooves alone.