Purpose: To compare splinted and individual restorations supported by short implants featuring an internal connection utilizing a split-mouth design. Materials and Methods: Splinted and nonsplinted implant crowns were prospectively compared in 18 patients. After verifying the need for at least two consecutive implants bilaterally, computed tomography scans were made, virtual planning was done, and qualifying patients were enrolled. Implants were placed using a two-stage surgical approach. After 3 to 5 months, patients were randomly restored with splinted prostheses on their left or right side. Nonsplinted restorations were made for contralateral sides. Radiographs were taken at prostheses seating and yearly exams. Radiographic bone levels were analyzed and compared (SAS 9.4) to determine differences between splinted and nonsplinted implants. Complications such as screw loosening, screw breakage, or porcelain fracture were assessed at recalls. Results: Eighteen patients (9 men and 9 women) with an age range from 49 to 76 years (mean = 56 years), received ≥ 4 implants in symmetrical posterior locations. Implants (n = 82) ranged in length from 6 to 11 mm with 70 implants ≤ 9 mm and 38 implants = 6 mm. At the time of this report, 3-year examinations and bone level comparisons were completed on 15 patients. One patient was lost to follow-up, one deviated from study protocol by smoking, and one was splinted on both sides due to repeated screw breakage. Screw loosening occurred in five patients on their nonsplinted side. These were 6-mm implants except for one patient. Porcelain chipping occurred for one patient on the splinted side. One 6-mm-length nonsplinted implant was lost after loading; this implant was successfully replaced after grafting. This patient had a total of six implants placed; ongoing bone level measurements included two pairs of implants only. For all implants combined, there was no significant difference (P > .05) at 1, 2, or 3 years for mean bone change around splinted and nonsplinted implants. However, length was identified as a significant factor (P = .0039). Further analysis revealed statistically significant differences between splinted and nonsplinted for 6-mmlength implants at 24 (P = .0061) and 36 (P = .0144) months. A gain in mean bone level of 0.41 and 0.37 mm was observed for nonsplinted implants at 24 and 36 months compared with baseline. Bone levels for the splinted 6-mm implants were not statistically different from baseline measurements (P > .05). Conclusion: Results of this prospective 3-year study of splinted ipsilateral and nonsplinted contralateral implants in 15 patients show: (1) peri-implant bone levels around splinted and nonsplinted implants were not statistically different for implants greater than 6 mm in length; (2) nonsplinted 6-mm implants revealed a gain in bone at 24 and 36 months compared with baseline; (3) all screw loosening only occurred on the nonsplinted side for 5 of 15 patients; and (4) implant loss after loading occurred for one 6-mm nonsplinted implant. Schlagwörter: nonsplinted prostheses, radiographic bone levels, splinted