Purpose: To show the influence of retentive cavity, cavity wall preparation and different luting techniques on the fracture resistance of severely damaged teeth restored with adhesive core build-ups and monolithic zirconium crowns. Methods: Extracted molars were prepared with 2 mm ferrule height and divided into eleven groups (n = 8/group). In nine groups a retentive occlusal cavity with a width and depth of 1 or 2 mm was prepared. Two control groups without a retentive cavity were made. Zirconium crowns were manufactured. 48 copings were cemented with glass-ionomer cement (Ketac Cem), the others (n = 40) with adhesive resin cement (Panavia F 2.0). Artificial ageing was carried out in the following way: n = 88, thermocycling (10,000 cycles, 6° C/60° C), n = 80 chewing simulation (1,200,000 cycles, 64 N). The samples were tested for load at first damage and fracture load with non-axial force. For statistical analysis ANCOVA with post hoc, Bonferroni-adjusted t-test were used ( p ≤ 0.05). Results: No differences between the tested cements were detected. Influence of the cavity wall thickness was significant ( p = 0.001). Mostly, the samples with wall thickness of 2 mm showed better results. Both control groups (no cavity) showed results comparable to study groups with cavity. Conclusions: Retentive cavity is most likely not mandatory. However, if prepared, the cavity wall thickness is of higher importance than cavity depth. Glass-ionomer and adhesive resin cement are comparable for use with zirconia crowns.