Purpose: To evaluate whether 4.0 mm short dental implants could be an alternative to augmentation with xenographs in the maxilla and placement of at least 10.0 mm long implants in posterior atrophic jaws. Materials and methods: A group of 40 patients with atrophic posterior (premolar and molar areas) mandibles with 5.0 mm to 6.0 mm bone height above the mandibular canal and 40 patients with atrophic maxillas having 4.0 mm to 5.0 mm below the maxillary sinus, were randomised according to a parallel group design to receive between one and three 4.0 mm long implants or one to three implants of at least 10.0 mm long in augmented bone, at two centres. All implants had a diameter of 4.0 mm or 4.5 mm. Mandibles were vertically augmented with inter-positional equine bone blocks and resorbable barriers. Implants were placed 4 months after the inter-positional grafting. Maxillary sinuses were augmented with particulated porcine bone via a lateral window covered with resorbable barriers, and implants were placed simultaneously. Implants were not submerged and were loaded after 4 months with provisional screw-retained reinforced acrylic restorations replaced after another 4 months by definitive screw-retained metal-composite prostheses. Patients were followed up to 1 year post-loading. Outcome measures were: prosthesis and implant failures, any complication, and peri-implant marginal bone level changes. Results: Three patients dropped out; one from the maxillary augmented group, one from the mandibular augmented group, and one from the maxillary short implant group. In six augmented mandibles (30%) it was not possible to place implants of at least 10.0 mm, so shorter implants were placed instead. In mandibles, one implant from the augmented group failed vs two 4.0 mm implants in two patients from the short implant group. In maxillae, three short implants failed in two patients vs seven long implants in four patients (two long implants and one short implant dropped into the maxillary sinus). Two prostheses on short implants (one mandibular and one maxillary) were placed at a later stage because of implant failures, vs six prostheses (one mandibular and five maxillary) at augmented sites (one mandibular prosthesis not delivered, three maxillary prostheses delivered with delays, one not delivered, and one failed) at augmented sites. In particular, three patients in the augmented group (one mandible and two maxillae) were not wearing a prosthesis. There were no statistically significant differences in implant failures (P (chi-square test) = 0.693; difference in proportion = 0.03; CI 95% -0.11 to 0.17) or prostheses failures (P (chi-square test) = 0.126; difference in proportion = 0.10; CI 95% -0.03 to 0.24). At mandibular sites, nine augmented patients were affected by complications vs two patients treated with short implants (P (chi-square test) = 0.01; difference in proportion = 0.37; CI 95% 0.11 to 0.63), the difference being statistically significant. No significant differences were found for maxillae: nine sinus-lifted patients vs four short implant patients were affected by complications (P (chi-square test) = 0.091; difference in proportion = 0.25; CI 95% -0.03 to 0.53). At 1-year post-loading, average peri-implant bone loss was 0.51 mm at 4 mm long mandibular implants, 0.77 mm at 10 mm or longer mandibular implants, 0.63 mm at short maxillary implants and 0.72 mm at long maxillary implants. The difference was statistically significant in mandibles (mean difference -0.26 mm, 95% CI -0.39 to -0.13, P (ANCOVA) < 0.001), but not in maxillae (mean difference -0.09 mm, 95% CI -0.24 to 0.05, P (ANCOVA) = 0.196). Conclusions: One year after loading 4.0 mm long implants achieved similar results, if not better, than longer implants in augmented jaws, but were affected by fewer complications. Short implants might be a preferable choice over bone augmentation, especially in mandibles, since the treatment is less invasive, faster, cheaper, and associated with less morbidity. However, 5 to 10 years post-loading data are necessary before making reliable recommendations.