Purpose: To evaluate whether immediate non-occlusal loading of single zirconia implants could reduce early failures when compared to immediate occlusal loading. Materials and methods: Forty partially edentulous patients who received one single zirconia implant (Z-Systems) at least 10 mm long and 3.25 mm wide inserted with a torque of at least 35 Ncm were randomised to immediate occlusal or non-occlusal loading groups. All patients received provisional acrylic crowns the same day of implant placement. Provisional crowns were replaced after 4 to 5 months by definitive full ceramic crowns. Outcome measures were implant success, any complications and peri-implant marginal bone levels. Results: One year after loading, no patients had dropped out. Five implants (12.5%) failed early: three occlusally loaded and two non-occlusally loaded. Three complications occurred, all after delivery of the definitive crowns: one crown fractured (occlusal loading), one had to be remade after debridement because of hyperplastic tissues (occlusal loading), and one crown decemented (nonocclusal loading). These differences were not statistically significant. Both groups gradually lost periimplant bone in a highly statistically significant way. One year after loading, patients subjected to non-occlusal loading lost an average of 0.7 mm of peri-implant bone versus 0.9 mm in the occlusal group. This difference in bone loss between groups was not statistically significant. There was an association between immediate post-extractive implants and implant failures (P = 0.01). Four of the 10 immediate post-extractive implants (40%) failed versus one out of 30 delayed implants (3%). Conclusions: The results of this study do not provide a conclusive answer to whether immediate non-occlusal loading may decrease implant failures. Immediately loaded zirconia implants placed in post-extractive sites had high failure rates. Keywords: immediate loading, non-occlusal loading, single implant, zirconia