Purpose: To evaluate the influence of at least three abutment disconnections in conventional loaded implants against placement of a definitive abutment in immediately non-occlusal loaded implants on hard and soft tissue changes. A secondary aim was to evaluate whether the presence of less than 2 mm of keratinised mucosa is associated with increased peri-implant marginal bone loss and soft tissue recessions. Materials and methods: Eighty patients requiring one single crown or one fixed partial prosthesis supported by a maximum of three implants were randomised, after implants were placed with more than 35 Ncm, according to a parallel group design to receive definitive abutments that were loaded immediately (definitive abutment or immediate loading group) or transmucosal abutments, which were delayed loaded after 3 months and removed at least three times: 1. At impression taking (3 months after implant placement); 2. When checking the zirconium core on titanium abutments at single crowns or the fitting the metal structure at prostheses supported by multiple implants; 3. At delivery of the definitive prostheses (repeated disconnection or conventional loading group). Patients were treated at four centres and each patient contributed to the study, with only one prosthesis followed for 3 years after initial loading. Outcome measures were: prosthesis failures, implant failures, complications, pink aesthetic score (PES), buccal recessions, patient satisfaction, peri-implant marginal bone level changes and height of the keratinised mucosa. Results: Forty patients were randomly allocated to each group according to a parallel group design. Six patients from the definitive abutment group dropped out or died, and one left from the repeated disconnection group. One implant, from the repeated disconnection group, fractured (difference = 3%; CI 95%: -2%, 8%; P = 1). Four provisional crowns and one definitive single crown had to be remade because of poor fitting, and one definitive crown and one definitive prosthesis because of ceramic and implant fracture, respectively, in the repeated disconnection group vs one provisional prosthesis from the definitive abutment group due to frequent debondings (difference = 15%; CI 95%: 2%, 28%; P = 0.060). Five patients from the definitive abutment group and four patients from the repeated disconnection group were affected by complications (difference = 4%; CI 95%: -11%, 20%; P = 0.725). PES scores assessed at 3 years post-loading were 11.7 (standard deviation = 1.8) mm for the definitive abutment group and 11.3 (1.5) mm for the repeated abutment changes group (difference = 0.4; CI 95%: -0.4, 1.2; P = 0.315). However, there was a difference of 0.26 out of a maximum score of 2 in favour of the definitive abutment group for soft tissue contour only. Buccal recessions at 3 years post-loading amounted to -0.1 (0.8) mm for the definitive abutment group and -0.1 (1.2) mm for the repeated abutment changes group (it was actually a soft tissue gain; difference = 0.01 mm CI 95%: -0.48, 0.50; P = 0.965). All patients declared being very satisfied or satisfied with the function and aesthetics of the prostheses and said they would undergo the same procedure again, with the exception of one patient from the repeated disconnection group who was uncertain regarding function. Mean peri-implant marginal bone loss 3 years after loading was 0.07 (0.18) mm for the definitive abutment group and 0.50 (0.93) mm for the repeated abutment changes group (difference = 0.43 mm; CI 95%: 0.13, 0.74; P = 0.007). The height of keratinised mucosa at 3 years post-loading was 2.8 (1.3) mm for the definitive abutment group and 2.8 (1.6) mm for the repeated abutment changes group (difference = 0.03; CI 95%: -0.67, 0.73; P = .926). Up to 3 years after initial loading there were no statistically significant differences between the two procedures, with the exception of 0.4 mm more marginal bone loss at implants subjected to three abutment disconnections. There were no significantly increased marginal bone loss (difference = 0.1 mm, CI 95%: -0.3, 0.5, P = 0.590) or buccal recessions (difference = 0.1 mm, CI 95%: -0.4, 0.7, P = 0.674) at implants with less than 2 mm of keratinised mucosa at loading. Conclusions: Three-year post-loading data showed that repeated abutment disconnections significantly increased bone loss of 0.43 mm, but this difference may not be considered clinically relevant; therefore clinicians can use the procedure they find more convenient for each specific patient. Immediately non-occlusally loaded dental implants are a viable alternative to conventional loading and no increased bone loss or buccal recessions were noticed at implants with less than 2 mm of keratinised mucosa.