Objective: To compare enamel and dentin microleakage of three single-bottle dentin-bonding agents (DBA) (One-Step, Single Bond, and Dentamed P&B) on teeth that were then subjected to thermocycling and occlusal loading. Method and materials: Class 5 cavities were prepared on the buccal and lingual surfaces of 30 extracted human molars and divided into three groups. The occlusal margins were in enamel, and the cervical margins were in cementum. DBA were applied strictly according to the manufacturers' recommendations, and the preparations were restored with resin composite. Specimens were subjected to 2,000 intermittent occlusal loads and 3,000 thermocycles and then immersed in 0.5% basic fuchsin dye for 72 hours. Four consecutive sections were cut for each specimen and examined under a stereomicroscope. The extent of dye penetration was measured and recorded using a nonparametric scale from 0 to 4. Results: The Mann-Whitney U test showed a significant difference at cementum margins between One-Step and Dentamed P&B, as well as between Single Bond and Dentamed P&B, and an insignificant difference between Single Bond and One-Step. There was no significant difference in dye penetration at enamel margins between the three DBA. Conclusion: Under combined use of occlusal loads and thermocycling, the sealing ability of single-bottle DBA at dentin margins ranged from good for One-Step to moderate for Single Bond to poor for Dentamed P&B. At enamel margins, all materials performed equally well. Keywords: dentin-bonding agents, microleakage, occlusal loading, thermocycling