The purpose of this in vitro study was to evaluate the microleakage of indirect composite inlays and to compare leakage of inlays with directly placed composite restorations. Standardized MOD preparations were cut in 50 extracted human molars. One gingival margin was placed in enamel above the CEJ, and the other was placed in dentin below the CEJ. Two groups of teeth were directly filled with composites (P-30 and Heliomolar) after being etched, and dentin bonding agents were applied. Two groups of teeth were restored with composite inlays that were fabricated on stone dies. The inlays were made and luted with the same two composites. The last group of teeth was restored with Heliomolar inlays luted with Dual cement. The specimens were thermocycled 300 times between 5 and 50 degrees C. Microleakage was evaluated by use of the silver-nitrate staining technique. The depth of leakage was measured microscopically after the teeth were sectioned. Both direct restorations and inlays showed substantial leakage at gingival-dentin margins; however, there was only superficial leakage at enamel margins. P-30 inlays and Heliomolar inlays cemented with Dual leaked less than direct restorations at the gingival-dentin margins. There was no difference in leakage of enamel margins of inlays and direct restorations, except that direct Heliomolar restorations leaked more than the others. There was no difference in leakage between Heliomolar restorations luted with light-cured or dual-cured cement.