The aim of this study was to investigate the interaction effect of different designs and surface treatments on the load-to-failure of lithium disilicate glass-ceramic repaired with nanofilled composite. Lithium-disilicate glass-ceramic slabs (IPS e.max Press, Ivoclar Vivadent) with three different designs of the top surface (flat, single plateau, or doubleplateau) (n=U) received 'no treatment', '5% HF etching', or "AI2O3 sandblasting". HF-etched and sandblasted slabs also received silane and universal one-step adhesive application. All slabs were incrementally repaired with nanofilled composite (Filtek Z350, 3M ESPE) up to6 mm above the highest ceramic top plateau. Specimens were stored in artificial saliva at 37 °C for 21 days and then subjected to 1,000 thermocycles between 5 and 55 °C. The interface composite-ceramic of each specimen was tensile tested until failure in a universal testing machine and the mode of failure was determined under a stereomicroscope. The ceramic surface morphology of one representative tested specimen from each subgroup (design/surface treatment) was observed through scanning electron microscopy (SEM). Regardless of ceramic design, the absence of surface treatment resulted in significantly lower load-to-failure values. No significant differences in load-to-failure values were observed between HF-etched and sandblasted specimens for the flat design; however, HF etching resulted in significantly higher load-to-failure values than sandblasting for both single plateau and double plateau designs. The majority (60%) of HF-etched specimens with single plateau or double plateau presented mixed failures. SEM photomicrographs showed that HF-etched specimens had smoother surfaces than sandblasted specimens. The surface treatment of a defective lithium disilicate glass-ceramic restoration has more influence than its macroscopic design on the retention of the composite repair. HF etching seems to provide higher bond strength to the composite repair.