This study assessed the marginal integrity and microleakage of standardized Class I resin composite restorations when placed with either "total" or "selective bonding" techniques. Sixty standardized Class I cavities comprising the main fissure system were prepared (9 mm length and 7 mm width). Cavity depth was set at 2.5 mm. In cavities where a glass ionomer liner was placed, the cavity was deepened by an additional 0.5 mm. In "total bonding" specimens, enamel and dentin were conditioned using a four-step adhesive system (Syntac Classic). In teeth with "selective bonding," a chemical curing conventional glass ionomer cement (GIC; Ketac Fil) and light-curing resin-modified glass ionomer liner (RMGI; Vitrebond) or three-step adhesive bonding liner (Syntac) were applied. The cavity margins of the latter specimens were finished with water-spray, acid-etched and a bonding agent was applied. All restorations were placed in two oblique increments. Totally bonded ceramic inlays (Cerec) served as the control. All specimens were subjected to thermo-mechanical loading (1.2 Mio cycles) and marginal quality and microleakage were assessed. The highest percentage of margins rated as "perfect" was found in selective bonding samples with glass ionomer liners and totally bonded inlay restorations. All the other groups showed significantly decreased marginal quality (p < 0.05). The same results were found for the microleakage assessment. The authors of the current study concluded that the application of a GIC liner significantly improved the overall marginal adaptation of direct Class I restorations. The use of an adhesive bonding agent for cavity sealing as currently used is not recommended.