Purpose: To investigate in vivo and in vitro Class II composite restorations performed with two matrix and wedge systems. Materials and Methods: One hundred nine Class II restorations were performed in 23 patients, 59 with metallic matrices and wooden wedges (group 1) and 50 with polyester matrices and reflective wedges (group 2). All cavities were restored using Single Bond and P-60 (3M ESPE). In the metal matrix group, polymerization was performed from the occlusal, and in the polyester group through the reflective wedge. To assess microleakage, 40 proximal standard slot cavities were prepared in 20 noncarious human third molars. In the mesial cavity, the gingival margin was located at the enamel level, and in the distal cavity at the cementum/ dentin. Specimens were randomly divided into two groups (n = 20) and restored with Single Bond, Z-250 (3M ESPE), and the same techniques used in the in vivo study: metal matrix/wooden wedge (group 1) and polyester matrix/reflective wedge (group 2). Specimens were thermocycled (500 times, 5°C to 55°C), then isolated with nail varnish and immersed in fuchsin for 8 h. Specimens were sectioned longitudinally and microleakage was assessed under magnification (40X) using a standard scoring system. Results: Data were subjected to the chi-square test, Mann-Whitney U-test, and Kruskal-Wallis H-test. Matrix systems presented similar results in the clinical evaluation and the in vitro microleakage test of Class II composite restorations. Dye leakage was minimal at enamel margins, and statistically lower (p 0.05) than at cementum/dentin margins. Conclusion: The different matrix systems had no influence on clinical performance or in vitro sealing ability of Class II composite restorations. Palabras clave: dental leakage, Class II restorations, matrix bands, clinical trials