Loss of teeth can affect a person’s self-esteem, social life, appearance and oral function. Reconstruction of a missing tooth has scientifically been shown to increase self-esteem and quality of life and to maintain oral function. For many patients a fixed dental prostheses (FDP) is preferred, either tooth- or implant-supported. Improvement and development of all-ceramic materials have made them preferable to other alternatives. However, despite properties of dental ceramics’ well known biocompatibility, good chemical and mechanical, the materials have their weaknesses, such as brittleness and some difficulties with the layering porcelain. Many all-ceramic materials cannot withstand minor flexure; more than 0.1 - 0.3 %, will lead to fracture. Oxide-ceramic, specifically yttria stabilized tetragonal zirconia polycrystals (Y-TZP) has become the most commonly used all-ceramic material. This material has the potential to be used for larger restorations. In addition, one of many challenges is to ensure durable zirconia-based restorations in the oral cavity. In the clinical situation, crowns and bridges are supported by a combination of different structures with differing properties, i.e. bone, dentine and enamel. The complexity of the supporting tissues in the oral cavity creates stress patterns in the prosthetic material, which need to be considered when designing a dental restoration. The durability of all-ceramic FDPs is dependent on knowledge of the material and design of the FDPs. In particular the design, shape of the connector and the radius of curvature at the gingival embrasure play a significant role in the load-bearing capacity of FDPs. The overall aim of this thesis is to evaluate design of zirconia-based restorations in relation to achieving increased fracture resistance. Another aim is related to how the choice of material used for supporting tooth analogues in the test set-up and how this influences test results relating to fracture strength of all-ceramic FDPs. Study I evaluates different radii (0.60 and 0.90 mm) of curvature in the embrasure of the connector area and different connector dimensions (2 x 2, 3 x 2 and 3 x 3 mm) and their effects on the fracture resistance of 3-unit all-ceramic FPDs made of Y-TZP. The results show that by increasing the radius of the gingival embrasure from 0.6 to 0.9 mm, the fracture strength for a Y-TZP FPD with connector dimension 3 x 3 mm will increase by 20%. Study II investigated how the choice of material (aluminium, polymer and DuraLay) used for supporting tooth analogues and support complexity influence test results concerning the fracture strength of FDPs made of a brittle material Y-TZP. The outcome of the study demonstrated that Y-TZP FDPs cemented on tooth analogues made of aluminium, with high E-modulus showed a significantly higher load at fracture and a different fracture mode than shown in clinical situations. Study III evaluates how factors as different default settings for connector design of two different CAD/CAM systems and different radii of curvature in the embrasure area of the connector will affect the fracture strength and the fracture mode of 3-unit, i.e. 4-unit allceramic FDPs made from Y-TZP and further to investigate how the number of pontics affect the fracture strength of Y-TZP. The results showed that the most crucial factor for the load-bearing capacity is the design of the radius of the gingival embrasures. Increasing the number of pontics from three to four decreases the load-bearing capacity nearly twice. Study IV investigate and compare the fracture strength and fracture mode in 11 groups of the currently most used multilayer all-ceramic systems for Y-TZP FDPs, with respect to the choice of core material, veneering material area, manufacturing technique (split-file, overpress, built-up porcelains and glass-ceramics), design of connectors and radius of curvature of FDP cores. The results show that the design of a framework is a crucial factor for the load bearing capacity of an all-ceramic FDP. The state-of-the-art designs are preferable, since the split-file designed cores call for a cross-sectional connector area, at least 42% larger, to have the same load bearing capacity as the state-of-the-art designed cores. Analyses of the fracture patterns demonstrated differences between the milled veneers and over-pressed or built-up veneers, where the milled ones showed numerically more veneer cracks whereas the other groups only showed complete connector fractures. All veneering materials/techniques tested were found, with great safety margin to be sufficient for clinical use both anteriorly and posteriorly.