OBJECTIVE: To evaluate in vitro microleakage of class V restorations using two self-etching primers with a flowable resin or hybrid resins. METHODS: Forty human molars were divided into 3 groups according to axial surfaces. Each group was randomly assigned to 2 subgroups (n=20) with either butt-joint or the beveled preparations. Class V preparations were cut in cemento-enamel junction. Groups A, B, C were respectively restored with CLB2/Clearfil AP-X, APMB/FHC-Merz or APMB/Liquicoat. Half specimens of each subgroup were thermocycled 2 500 times. After staining, dye penetration was evaluated in the ordinal scale at 40Xmagnifications for occlusal and the gingival margins. The wall adaptation of the randomly selected specimens was analyzed with a SEM on replicas of the sectioned teeth. RESULTS: None of bond systems in this study prevented the microleakge. The restorations with the bur-beveled preparations leaked the same as those with the butt joint preparations (P>0.05). The thermocycled specimens and the non-thermocycled specimens leaked similarly (P>0.05). The flowable resin-liqucoat leaked significantly more than hybrid resin in dentinal margins (P<0.05). CONCLUSION: The self-etching primers can not reduce the microleakge in the dentin-restoration interfaces. The flowable resin leak more than the hybrid resin in dentin-restoration interfaces. The bur- beveled preparations do not significantly reduce the microleakage in class V resin restorations bonded with self-etching primers.