Purpose: To compare the clinical outcomes of immediate versus conventional delayed loading of four dental implants in edentulous mandibles with fixed prostheses. Materials and methods: A blinded, two-arm, parallel group, randomised controlled trial was conducted. A total of 42 patients were included, and each received four Brånemark System implants with a TiUnite surface. The patients were randomly assigned to two study arms: 1) immediate-loading arm (IL), in which the mandibular denture was converted into an interim implant-supported fixed prosthesis (ISFP) on the day of surgery, with a permanent ISFP being inserted at least 3 months postsurgery; 2) conventional-loading arm (CL), in which the mandibular removable prosthesis was relieved at the implant site and relined with a soft tissue conditioner. Only implants with a minimum insertion torque of ≥ 35 Ncm were included in the IL group. Implants were loaded 4 to 6 months postsurgery. Independent, blind investigators assessed the patients at 2, 6 and 12 months and at 10 years. The outcome measures were prosthesis and implant success rates, type and frequency of complications and changes in peri-implant marginal bone levels. Results: A total of 20 patients were allocated to the IL group and 22 to the CL group. However, one patient from the IL arm was excluded and three patients were reallocated to the CL arm. Two implants in one patient and one in another patient could not be placed with a ≥ 35 Ncm insertion torque, and a third patient developed severe sudden gag reflex and thus it was not possible to load the implants immediately. At a later stage, one of the patients who failed the initial stability test dropped out of the study. Therefore, initially, 24 patients were conventionally loaded and 16 patients were immediately loaded. At the 10-year follow-up, six patients dropped out from the IL arm and two from the CL arm. Also, at the 10-year-follow-up, the CL and IL study arms consisted of 22 and 10 participants, respectively, using the per-protocol (PP) analysis. Six implants failed in two patients of the CL arm (two implants in one patient and four implants in another patient), and three implants failed in three patients in the IL arm (PP analysis), respectively. The patient-level implant failure rate was 10% (intention-to-treat [ITT] analysis) and 14% (PP analysis) in the CL arm, and 25% (ITT) and 20% (PP) in the IL arm. The difference was not statistically significant (95% CI from −0.18 to 0.39, P = 0.65). The failure rate at the implant level was 8% (ITT) and 8% (PP) in the CL arm, and 6% (ITT) and 5% (PP) in the IL arm. The difference was not statistically significant (95% CI from −0.06 to 0.14, P = 0.44). Ten years after loading, patients in the IL arm lost an average of 0.55 ± 0.64 mm of peri-implant bone versus 0.41 ± 0.40 mm of peri-implant bone loss observed in the CL arm. The 10-year bone loss in both arms was statistically significant compared with the baseline (P 0.001). However, there was no statistically significant difference between the two arms for peri-implant bone level changes (the difference between the arms was 0.14 mm ± 0.50 mm; 95% CI -0.23 to 0.52; P = 0.43). One prosthesis failed due to the loss of all four implants in one patient of the CL arm. Eight patients from the IL arm were affected by 13 complications (such as pain from fractures and inflammation) versus seven patients (10 complications) from the CL arm. The complication rate was 67% in the IL arm and 35% in the CL arm. The difference in complication proportions between the two arms was not statistically significant (difference in proportions = 0.32; 95% CI = -0.08 to 0.61; P = 0.14). All complications were managed successfully. Conclusions: Long-term data of immediate loading of four dental implants with a mandibular fixed prosthesis revealed comparable clinical outcomes to conventional loading. Therefore, immediate loading should be considered in the treatment of edentulous patients. Keywords: dental implants, immediate loading, mandible, prosthesis Conflict of interest statement: The study was supported by the Nobel Biocare Graduate Students Grant, Gothenburg, Sweden.