Objective: The aim of this study was to determine the microleakage of current-generation dentin bonding systems in Class II resin composite restorations. Method and materials: Class II (occlusodistal or occlusomesial) cavity preparations with a gingival margin 2 mm apical to the cementoenamel junction were prepared on 70 noncarious, extracted human premolar teeth. Teeth were randomly divided into five groups and treated with different-generation bonding systems (Optibond FL, Gluma One Bond, Clearfil SE Bond, acid etching plus Clearfil SE Bond, and Prompt-L-Pop). All cavities were restored in a posterior resin composite and subjected to 200 thermocycles (at 5°C to 55°C; 30-second dwell time). After immersion in 0.5% basic fuchsin dye for 24 hours, the teeth were sectioned longitudinally and evaluated for dye penetration using a binocular stereomicroscope. Results: There was no statistically significant differences among bonding systems in the degree of microleakage in the occlusal walls. For gingival walls, statistically significant differences were found only between the Clearfil SE Bond and Prompt-L-Pop groups and the Clearfil SE Bond with acid etching and Prompt-L-Pop groups. The greatest microleakage was observed in Prompt-L-Pop specimens. Conclusion: Most of the tested dentin bonding systems were able to eliminate microleakage completely in the occlusal walls, but some systems exhibited statistically significant differences in leakage in the gingival walls.