To evaluate the microleakage of Cerec 3, IPS e.max Press, and Turkom-Cera inlays cemented with three self-adhesive resin cements. Ninety standardized class III MOD cavities were prepared in intact human mandibular third molars. Ceramic inlays were fabricated according to the manufacturer's instructions and were cemented using three self-adhesive resin cements (RelyX Unicem, Smartcem 2, and SpeedCEM). The specimens were stored in distilled water at 37°C for 24 h and subjected to 1000 thermocycles in water between 5°C and 55°C with a dwell time of 30 s. Subsequently, the specimens were subjected to 100,000 cycles of mechanical loading of 50 N at 1.6 Hz in 37°C water. The specimens were immersed in 0.5% basic fuchsine for 24 h and were sectioned using a low-speed diamond blade. The percentage of dye leakage at the tooth/restoration interface was measured and compared by Kruskal-Wallis tests with Bonferonni correction and Mann-Whitney U-tests at a significance level of p<0.05. Microleakage at the RelyX Unicem interface was lower than that with Smartcem 2 and SpeedCEM resin cements (p<0.05). Microleakage of the Turkom-Cera system was higher than Cerec 3 and IPS e.max Press ceramic inlays (p<0.05). Regardless of the ceramic system and self-adhesive resin cement used, dentin margins were associated with higher microleakage than enamel margins.