Objectives: To compare the effectiveness of onlay bone blocks of equine origin (test or XB group) with autogenous bone blocks (control or AB group) harvested from the ramus or the iliac crest for the rehabilitation of partially or fully edentulous atrophic jaws with implant supported prostheses. Materials and methods: Forty patients with partially or fully edentulous atrophic jaws having less than 5 mm of residual crestal bone height and/or less than 3 mm of bone thickness, as measured on computerised tomography (CT) scans, were randomised into two groups according to a parallel group design, either to be augmented with autogenous onlay bone blocks (20 patients; AB group) from the mandibular ramus or the iliac crest, or with onlays blocks of spongious bone of equine origin (20 patients; XB group). Two centres treated 20 patients each. Six XB blocks were modelled on lithographic models of the jaws before grafting. The blocks were fixed with screws and osteosynthesis plates and were covered with resorbable barriers made of equine cortical bone and fixed with tacks. The autogenous bone grafts were left to heal for 4 months and the xenografts for 7 months before placing implants, which were submerged. After 4 months, either bar-retained overdentures or provisional reinforced acrylic prostheses were delivered. Provisional prostheses were replaced, after 4 months, by definitive fixed prostheses. Outcome measures were: prosthesis and implant failures; complications; patient satisfaction; pain recorded 3 and 10 days post-augmentation; number of days of hospitalisation, total and partial infirmity days. All patients were followed for 4 months after loading. Results: All patients could be rehabilitated with implant-supported prostheses and none dropped out. Twenty-eight patients were augmented in the maxilla (15 with AB and 13 with XB) and 12 in the mandible (5 with AB and 7 with XB). No AB graft failed totally versus 10 XB grafts (difference = 0.5; 95% CI 0.23 to 0.68; P = 0.0004). In particular, all 7 XB mandibular grafts and 5 out of 6 XB blocks (3 in mandibles and 2 in maxillas), which were previously modelled on lithographic models of the jaws failed. One implant failed in one AB patient versus 11 implants in 4 XB patients (P = 0.3416). All but 1 prostheses were loaded in time in the AB patients, versus 4 prostheses which were loaded with delays in XB patients because of graft and implant failures (P = 0.3416). Four complications occurred in 4 AB patients versus 15 complications in 12 XB patients (difference = 0.4; 95% CI 0.09 to 0.63; P = 0.0225). Fourteen AB patients reported moderate pain 3 days postoperatively versus 6 XB patients (P = 0.0562); at 10 days, 10 AB patients reported moderate pain versus 1 XB patient (difference = -0.45; 95% CI -0.65 to -0.17; P = 0.0033). The 14 patients harvested from the iliac crest were hospitalised for an average of 3.1 nights, whereas 7 patients treated with XB were hospitalised on average for 1.4 nights (P 0.0001). The number of total and partial infirmity days was 126 for theAB group and 43 for the XB group, and 220 for the AB group and 93 for the XB group, respectively (mean day difference = -4.15; 95% CI -7.35 to -0.95; P = 0.0134 and mean day difference = -5.7; 95% CI -10.01 to -1.39; P = 0.0116, respectively). Seventeen AB patients versus 19 XB patients were fully satisfied with function of their prostheses (P = 0.6050), 18 AB patients versus 12 XB patients were fully satisfied with aesthetics of their prostheses (P = 0.0648), and 5 and 3 patients, respectively would not undergo the same procedure again (P = 0.6948). There were no differences between the outcomes of the two centres with exception of prosthesis failures and complications in the maxilla. Conclusions: Autogenous onlay bone blocks are superior to equine onlay bone blocks, especially in mandibles, where all equine blocks failed, therefore we strongly discourage the use of onlay bone blocks of equine origin in mandibles. Schlagwörter: atrophic jaw, autogenous bone, bone substitutes, blocks, iliac crest Conflict of interest statement: Bioteck, Arcugnano (VI), Italy, partially supported this trial, however data belonged to the authors.