Incidence of using esthetic composite resin in the posterior area is increasing but there were lots of inconsistent reports about their microleakage and marginal adaptation. The purpose of this study was to evaluate the differences of microleakage and marginal adaptation according to restorative techniques. 30 cavities with enamel gingival margin were prepared and restored with 3 types of composite resin [Z-100($Scotchbond^{TM}$ MP), AELITEFIL ($Onestep^{TM}$), Her culite XRV(Fuji BOND LC)] in direct technique and another 30 cavities were restored with preformed CR inlays and 3 different modern resin and resin-modified GI cements (Superbond C&B, Choice, Fuji Duet). Samples were chemically stressed in 75% ethanol for 24 hours and were thermocycled (5-$55^{circ}C$(500 times. The degree of microleakage through proximal and gingival margins was examined by 1 % MB dye and the degree of marginal adaptation by examining the margins via SEM. The following results were obtained ; 1. In direct groups, Herculite XRV [Fuji BOND LC, 35.13 (15.50) %] group showed statistically different, less microleakage than Z-100 [$Scotchbond^{TM}$ MP, 72.91 (16.91 %] group and AELITEFIL [One-step, 93.73 (13.66) %] group (p<0.05). 2. In indirect groups, the degree of microleakage in Mean(S.D.) were: Super bond C & B [39.00 (24.35) %], Choice [57.19 (33.80) %], Fuji Duet [58.22 (40.36) %]. But there was no significant difference. 3. There was no significant difference between resin cement and resin-modified GI cement. 4. There were gaps at the interface with the tooth structure, but no gap was seen at the interface with restoration in all specimens. 5. In direct groups, Herculite XRV(Fuji BOND LC) group made little gap compared with other groups, but 40-$50{mu}m$ thickness of bonding agent, Fuji BOND LC, looked like a cement used in indirect technique. 6. All indirect groups showed a variety of cement thickness, from less than $20{mu}m$ to over $100{mu}m$ and that dimension of buccal/lingual margin was less than that of gingival margin.