To evaluate presumptive differences in osseointegration at implants supporting crowns that are physiologically loaded either immediately or 3 months after installation. All premolars and first molars were extracted bilaterally in six dogs. After 3 months of healing, three implants were installed on the premolar region and two in the molar region in one side of the mandible. Likewise, after another 3 months, five implants were installed in the contralateral side, and impressions were taken bilaterally. Within 48 hours, two single crowns were screwed bilaterally onto two implants in the premolar region, and two splinted crowns reproducing the shape of the first molar were screwed bilaterally onto the implants in the molar region. The mesial implants were used as no-loaded controls. Sacrifices were performed after 3 months, and histological analyses were performed. At the premolar sites, mineralised bone-to-implant contact (MBIC%) was 78.0 ± 4.0% and 70.9 ± 7.9% at the delayed and immediately loaded sites, respectively. This difference was statistically significant. At the control implants, MBIC% was 61.4 ± 14.7% and 63.1 ± 13.1% at the delayed and the immediately loaded sites, respectively. At the molar sites, MBIC% was 79.2 ± 10.9% and 61.1 ± 10.3% at the delayed and immediately loaded sites, respectively. Applying a delayed loading to fixed dental prostheses supported by single or two splinted implants yielded higher proportions of bone-to-implant contact (osseointegration) compared to immediately loaded implants. Moreover, both types of loading protocols yielded a higher rate of osseointegration compared to unloaded implant sites after 3 months following implant installation.