Restoration microleakage is thought to be determined by the method and location of cavity preparation, enamel etching, and dentin conditioning, as well as the restorative material. This study compared the microleakage of composite restorations placed in preparations cut with carbide and diamond burs and those treated with different bonding/conditioning agents. Class V preparations (3 x 2 x 2 mm) were cut wholly in enamel or in enamel and cementum in 100 human premolars. Twenty teeth were prepared with carbide burs and the preparations etched and conditioned with Prime&Bond NT but not restored. A second set of 20 teeth had enamel-only preparations cut with carbides (n = 10) or diamonds (n = 10), and the preparations etched, conditioned (Prime&Bond NT) and restored with Prisma TPH. The other 60 teeth were divided into 3 groups of 20 teeth each with enamel-only (n = 10) or enamel/cementum preparations (n = 10). The 3 groups of teeth were conditioned with Optibond Solo, Clearfil SE Bond or Prompt-L-Pop prior to restoration with Prisma TPH. Two mm of root was resected from all teeth, pulpal tissue removed, and insulated copper wires inserted via the root canals to contact with the pulp chamber roof before the tooth-wire interfaces and root surfaces were sealed. The teeth were immersed in 0.9% NaCl and leakage assessed over 30 d by iR drop across a resistor in series with a DC source and stainless-steel counter electrode. Differences (p 0.05) in leakage were found for enamel preparations cut with carbides and diamonds, and the relationship of leakage vs time was linear. Enamel/cementum preparations showed greater leakage, and the relationship of leakage vs time was sigmoidal. Conditioned-only preparations showed the same leakage as those conditioned and restored, while preparation leakage varied with the conditioning agent. Cavity preparation location, method of cutting, and the conditioning agent markedly affect leakage behavior. Schlagwörter: restoration leakage, cavity preparations, dental burs, dentin conditioning