The purpose of this study was to compare the polymerization shrinkage stress of composite resins (microfilled, microhybrid and hybrid) photoactivated by quartz-tungsten halogen light (QTH) and light-emitting diode (LED). Glass rods (5.0 mm x 5.0 cm) were fabricated and had one of the surfaces air-abraded with aluminum oxide and coated with a layer of an adhesive system, which was photoactivated with the QTH unit. The glass rods were vertically assembled, in pairs, to a universal testing machine and the composites were applied to the lower rod. The upper rod was placed closer, at 2 mm, and an extensometer was attached to the rods. The 20 composites were polymerized by either QTH (n=10) or LED (n=10) curing units. Polymerization was carried out using 2 devices positioned in opposite sides, which were simultaneously activated for 40 s. Shrinkage stress was analyzed twice: shortly after polymerization (t40s) and 10 min later (t10min). Data were analyzed statistically by 2-way ANOVA and Tukey's test (alpha=5%). The shrinkage stress for all composites was higher at t10min than at t40s, regardless of the activation source. Microfilled composite resins showed lower shrinkage stress values compared to the other composite resins. For the hybrid and microhybrid composite resins, the light source had no influence on the shrinkage stress, except for microfilled composite at t10min. It may be concluded that the composition of composite resins is the factor with the strongest influence on shrinkage stress.