Purpose: To evaluate preparation designs of compromised cusps and whether or not they influence masticatory fatigue, fracture resistance, and marginal discrepancy of ceramic partial-coverage restorations (PCRs) luted on mandibular molars. Materials and Methods: Sixty-four caries-free molars were equally divided into four groups. Control group NP received no preparation (NP). Group B-IN received a basic inlay (IN) preparation with buccal (B) cusp conservation and occlusal reduction of both lingual cusps. Group B-ON was prepared in the same way, except buccal cusps were prepared with an angle of 45 degrees to the occlusal plane (buccal onlay). Group B-OV preparation was similar to group B-ON, but buccal cusps received a further shoulder preparation on the buccal aspect (buccal overlap). Forty-eight all-ceramic IPS e.max Press PCRs were fabricated and luted adhesively. Specimens underwent mouth-motion fatigue (1.2 million cycles, 1.6 Hz, 49 N) and 5500 thermal cycles (5°C/55°C). Fracture patterns were observed. Surviving specimens were loaded until fracture. Marginal discrepancies were examined. Results: Only one specimen of group B-ON fractured during fatigue. Median fracture loads (N) [IQR=x.25-x.75]: group NP = 1604 N [1182-1851 N], group B-IN = 1307 N [1262-1587N], group B-ON = 1396 N [817-1750N], group B-OV = 1205 N [1096-1542N]. No significant differences in fracture resistance were found between restored molars and unprepared teeth (p >= 0.18). Different preparation designs showed no significant influence on PCR fracture resistance. Mouth-motion fatigue caused a significanty decrease of marginal accuracy in groups B-IN (p = 0.009) and B-ON (p = 0.008). Marginal discrepancy values of groups B-IN and B-OV were significantly different after fatigue (p = 0.045). Conclusion: Ceramic coverage of compromised cusps did not demonstrate an increase of fracture resistance after fatigue when compared to less invasive partial-coverage restorations. However, enhanced exposure of restoration margins to occlusal wear could result in more extensive marginal discrepancies. Schlagwörter: fracture resistance, marginal discrepancy, partial coverage restoration, ceramic, fatigue