This in vitro study evaluated microleakage in Class II cavities restored with dental composite and varying light-curing units and the temperature of the composite when subjected to a thermocycling test. Ninety cavities were prepared on the proximal surfaces of bovine teeth and randomly divided according to the light-curing mode (QTH-420 mW/cm2, LED 2nd generation-1100 mW/cm2, or LED 3rd generation-700 mW/cm2) and temperature of the resin composite (23°C, 54°C and 60°C). Following the restorative procedures and thermocycling, the samples were immersed in methylene blue for 12 hours. The samples were ground and the powder prepared for analysis in an absorbance spectrophotometer. All the results were statistically analyzed using the nonparametric tests of Kruskal-Wallis and Dunn (p ≤ 0.05). The results showed that there was no statistical difference between the light-curing modes at a temperature of 23°C. For 54°C, QTH showed a microleakage mean that was significantly lower than those of the LED groups, and for 60°C, QTH had a microleakage mean significantly lower than that of the LED 2nd generation group. There was no statistical difference between the temperatures of the resin composite when LEDs were used. For QTH, 54°C showed statistically lower microleakage than 23°C. The group preheated to 60°C showed no difference when compared to the group heated to 23°C. Preheating the resin composite (54°C and 60°C) did not improve the microleakage means when high-irradiance LED was used; however, it decreased the microleakage means when a QTH with low irradiance was used.