Cuspal deflection produced by polymerization shrinkage was measured after using different composite restoration techniques. This study included 30 healthy premolars embedded in acrylic resin connected to a system that simulated intrapulpal pressure. A small ball was attached to each cuspal vertex as a reference point for intercuspal distance measurements. A large mesio-occlusal cavity was cut in each premolar. All premolars were treated with the same adhesive (ScotchBond) and composite (Tetric Ceram). The teeth were randomly distributed among three study groups: Group 1, filled with two horizontal increments; Group 2, filled with two horizontal increments, the first up to half the cavity height and light cured using a transparent plastic cone (Cerana), and the second filling the remainder of the cavity and Group 3, filled in the same way as Group 2, except that the first increment only filled one-third of the cavity height. The intercuspal distance was measured before beginning the restoration and immediately after polymerization of the first and second increments. Under the experimental conditions used, none of the filling techniques utilized avoided the cuspal deflection phenomenon. Polymerization of the final increment, which binds occlusal enamel in the buccal-lingual plane, was the main cause of cuspal deflection and produced a statistically significant reduction in intercuspal distance vs the baseline measurement in the three study groups. The global deflection ranged from 4 microm to 6 microm, depending on the filling technique used, although the differences among techniques did not reach statistical significance (p<0.05).