The objective of this study was to investigate the effect of different storage media on microleakage at the gingival margin of mixed Class V composite restorations. Human molars were either used immediately after extraction (control group) or after a 3-4 wk storage period in 1% chloramine, 70% ethanol, 10% formalin, or 0.1% thymol. Class V cavities with cervical margins in cementum or dentin were prepared and restored with a microfill composite using five different dentin bonding agents. After thermocycling (1440 x from 5 degrees C to 55 degrees C) and basic fuchsin staining, the depth of dye penetration at the gingival margins was determined and ranked on a scale of 0 to 4. Class V restorations in teeth stored in formalin showed significantly less microleakage (mean leakage score, MLS = 1.7) than restorations in freshly extracted teeth (control group; MLS = 3.1). Microleakage scores in teeth stored in chloramine (MLS = 2.7), ethanol (MLS = 3.0) or thymol (MLS = 3.2) were not significantly different from the microleakage scores in the control group. The improvement of in vitro efficacy of dentin bonding agents when using extracted teeth stored in formalin is probably a result of collagen cross-linking by formaldehyde. Since such an alteration of the dentin does not reflect the condition of the substrate in vivo, extracted teeth assigned for in vitro evaluation of dentin bonding agents should not be stored in formalin. In teeth stored in 1% chloramine, the sums of ranks and the rank orders of the dentin bonding agents were comparable to those in freshly extracted teeth.(ABSTRACT TRUNCATED AT 250 WORDS)