We analyzed the effect of the crosshead speed of an applied load on failure load and failure mode of restored human premolars. Fifty intact, noncarious human premolars were selected. Class II mesio-occlusodistal preparations were made with a water-cooled high-speed preparation machine, and the teeth were restored with composite resin. The specimens were divided into five groups (n = 10 each) and tested individually in a mechanical testing machine, in which a 6.0-mm-diameter steel cylinder was mounted to vary the crosshead speed: v0.5: 0.5 mm/min; v1: 1.0 mm/min; v2.5: 2.5 mm/min; v5: 5.0 mm/min; and v10: 10.0 mm/min. The cylinder contacted the facial and lingual ridges beyond the margins of the restorations. Peak load to fracture was measured for each specimen (N). The means were calculated and analyzed with one-way analysis of variance followed by Tukey's test (a = 0.05). The mean load at failure values were (N) as follows: v0.5, 769.4 ± 174.8; v1, 645.2 ± 115.7; v5, 614.3 ± 126.0; v2.5, 609.2 ± 208.1; and v10, 432.5 ± 136.9. The fracture modes were recorded on the basis of the degree of the tooth structural and restorative damage: (I) fracture of the restoration involving a small portion of the tooth; (II) fractures involving the coronal portion of the tooth with cohesive failure of the composite resin; (III) oblique tooth and restoration fracture with periodontal involvement; and (IV) vertical root and coronal fracture. Varying crosshead speeds of 0.5–5.0 mm/min did not influence the failure load of restored maxillary premolars; however, increasing the crosshead speed to 10 mm/min decreased the failure load values and the degree of tooth structural damage.