OBJECTIVE To test the strength, at the marginal ridge, of the tunnel preparation vs the Class II traditional box preparation when restored with composite or glass ionomer (GI). METHODS Eighty-four extracted maxillary molars stored in normal saline and thymol were randomly divided into six groups of 14 each (determined by pilot study where alpha = 0.05, beta = 0.2). Group A and C were tunnel preparations. Group B and D were traditional Class II preparations. Group E-whole tooth (negative control), and Group F-tunnel preparation unrestored (positive control). Group A and B were restored with Scotchbond 2/P-50 composite. Group C and D were restored with Ketac Fil-GI. The marginal ridge of each tooth was loaded at 0.5 mm/minute on the Instron. A loading rod produced a contact point of 1.0 mm in diameter. RESULTS The mean compressive loads (kg) required for fracture were: (A): 42.2 +/- 11.9, (B): 53.1 +/- 10.7, (C): 52.0 +/- 10.9, (D): 23.8 +/- 8.4. (E): 79.1 +/- 16.1, (F): 27.0 +/- 10.6. A significant difference was found between whole teeth (E) and all other groups (P < 0.05). The Newman-Keuls test showed a significant difference between Class II composite (B) and tunnel composite (A) (P < 0.05), between tunnel GI (C) and tunnel composite (A) (P < 0.05) but no difference between tunnel GI (C) and Class II composite (B).