Purpose: To evaluate the outcome of two vs four implants placed flapless in fully edentulous mandibles and immediately restored with metal-resin screw-retained cross-arch prostheses. Materials and Methods: A total of 60 patients from two different centres were randomised: 30 to the fixed-on-2 group and 30 to the fixed-on-4 group according to a parallel group design. To be immediately loaded implants had to be inserted with a minimum torque of 40 Newton/cm (Ncm). Outcome measures were prosthesis and implant failures, complications, and marginal bone level changes. Results: Flaps were raised in 18 patients. Two implants in two patients did not reach the planned insertion torque and were replaced immediately by larger diameter ones. Three years after loading, two patients dropped out from the fixed-on-2 group and one from the fixed-on-4 group. No implant failure or prosthetic failure occurred. A total of 12 fixed-on-2 patients were affected by one complication each vs 13 fixed-on-4 patients. There were no statistically significant differences for complications between groups (difference in proportion = -0.03; 95% CI -29 to 0.22; P [Fisher's exact test] =1.000). There were no statistically significant differences for marginal peri-implant bone loss between the two groups (mean difference = 0.14; CI 95% -0.17 to 0.45; P [ANCOVA] = 0.356), with both groups losing marginal bone in a statistically significant way (1.70 mm for fixed-on-2 implants and 1.56 mm for fixed-on-4 implants). There were statistically significant differences between the two centres, with more bone being lost for the Bologna centre (2.18 vs 1.13 mm; P (ANCOVA) < 0.001). Conclusions: Up to 3 years after loading, these preliminary results suggest that immediately loaded mandibular cross-arch fixed prostheses can be supported by only two dental implants. Longer follow-ups (about 10 years) are needed to properly evaluate this therapeutic option.