PURPOSE. Zirconia materials have been used for implant-retained restorations, but the stress distribution of zirconia is not entirely clear. The aim of this study is to evaluate the stress distribution and risky areas caused by the different design of zirconia restorations on the atrophic bone of the posterior maxilla. MATERIALS AND METHODS. An edentulous D4-type bone model was prepared from radiography of an atrophic posterior maxilla. Monolithic zirconia and zirconia-fused porcelain implant-retained restorations were designed as splinted or non-splinted. 300-N occlusal forces were applied obliquely. Stress analyses were performed using a 3D FEA program. RESULTS. According to stress analysis, the bone between the 1) molar implant and the 2) premolar in the non-splinted monolithic zirconia restoration model was stated as the riskiest area. Similarly, the maximum von Mises stress value was detected on the bone of the non-splinted monolithic zirconia models. CONCLUSION. Splinting of implant-retained restorations can be more critical for monolithic zirconia than zirconia fused to porcelain for the longevity of the bone.