Using the push-out test, this study evaluated the bond strength of the composite resin Z250 (3M/ESPE) photoactivated with the XL2500 (3M/ESPE) quartz-tungsten-halogen light-curing unit with different curing protocols: soft-start (150 mW/cm² for 2 s (S2), 3 s (S3), 5 s (S5), 10 s (S10) or 15 s (S15), followed by 700 mW/cm² for 15 s; pulse-delay (150 mW/cm² for 2 s (P2), 3 s (P3), 5 s (P5), 10 s (P10) or 15 s (P15), with a 1-min delay, followed by 700 mW/cm² for 15 s. After storage at 37°C ± 1 for 24 h ± 1, the specimens were ground, polished and subjected to a push-out test in a universal test machine (Instron) with a cell load of 500 N at a crosshead speed of 0.5 mm/min. The data were analyzed statistically by ANOVA and Tukey's test at 5% significance level. There were no statistically significant differences (p>0.05) among the groups photoactivated using the soft-start mode. For the pulse-delay mode, P5 promoted the highest bond strength (p<0.05). Groups photoactivated with the pulse-delay mode (except for P2 and P15) presented significantly higher bond strength than those photoactivated with the soft-start. It may be concluded that the influence of initial exposure time was curing method-dependent, with the best results obtained using the pulse-delay method with 5 s in the first photoactivation cycle.