Purpose: To test the hypothesis that platform switching and regular platform implants would have different outcomes in single-tooth replacement against the alternative hypothesis of no difference. Material and methods: This study was designed as a randomised controlled split-mouth trial. Eighteen patients with bilaterally missing single premolars or molars to be restored with implant-supported single crowns, were consecutively enrolled. Implant sites were randomly assigned to be treated according to the platform switching concept (PS group), or with matching implant-abutment diameters (RP group). A total of 36 Nobel Replace Tapered Groovy implants were installed. All the implants were inserted in healed bone, with an insertion torque between 35 and 45 Ncm, according to a one-stage protocol. Both implant types were loaded with a screw-retained temporary crown 3 months after implant insertion. Definitive screw-retained single crowns were delivered 2 months later. Outcome measures were implant and prosthetic survival rates, biological and prosthetic complications, radiographic marginal bone level (MBL) changes, pocket probing depth (PPD) and bleeding on probing (BOP). Clinical data was collected at implant placement (baseline), and at 3, 9 and 36 months after loading. Results: No patients dropped out and no implant failed. No prosthetic complications were recorded. One patient experienced mucosal inflammation with positive BOP (RP group) after 3 months, three patients had bilateral peri-implant mucosal inflammation with positive BOP at 6, 24 and 30 months after loading, respectively. There were no statistically significant differences between groups for complications (3/18 versus 4/18; P = 1.0; Odds Ratio = 1.333; 95% CI: 0.3467 to 5.1272). Nine months after loading, the mean MBL was 0.93 ± 0.26 mm in the RP group and 0.84 ± 0.23 mm in the PS group, with no statistically significant differences between groups (mean difference = 0.09 mm, 95% CI: -0.22 to 0.04, P = 0.18). Three years after loading, mean MBL was 1.09 ± 0.31 mm in the RP group and 1.06 ± 0.24 mm in the PS group, with no statistically significant differences between groups (mean difference = 0.02 mm, 95% CI: -0.06 to 0.10, P = 0.70). Marginal bone level changes between 3 years and baseline were 0.72 ± 0.28 mm in the RP group and 0.71 ± 0.27 mm in the PS group, with no statistically significant differences between the groups (mean difference = -0.00 mm, 95% CI: -0.07 to 0.07, P = 0.89). Mean PPD was 2.70 ± 0.52 mm in the RP group and 2.46 ± 0.69 mm in the PS group at 36 months after loading, with no statistically significant differences between the groups (mean difference = 0.23 mm, 95% CI: -0.05 to 0.35, P = 0.43). Mean BOP was 0.83 ± 0.96 mm in the RP group and 0.89 ± 0.99 mm in the PS group at 36 months after loading, with no statistically significant differences between the groups (mean difference = 0.07 mm, 95% CI: -0.03 to 0.17, P = 0.77). Conclusions: The clinical and radiographic outcomes of implants restored according to the platform-switching concept versus implants restored with the matching implant-abutment diameters are comparable, 3 years after loading. Keywords: dental abutment, dental implants, implant-abutment interface, marginal bone loss, platform switching Conflict of interest statement: This study was not supported by any company. All authors declare no conflict of interest.