We inserted, in 8 dogs, implants with either porous-coated or grit-blasted titanium surface and coated with hydroxyapatite (HA) into trabecular bone in the proximal humerus, using a 1 mm gap model. After 25 weeks, push-out tests showed that energy absorption for porous-coated implants was twice that of grit-blasted implants, whereas shear stiffness was reduced by one fifth, indicating a stronger fixation of porous-coated implants. Macroscopically, all grit-blasted implants had delamination of the HA coating, whereas porous implants failed mostly at the HA-tissue interface. Porous-coated implants had 47% bone ingrowth and grit-blasted implants 70% (p = 0.02), however, no difference in absolute surface area was found. Part of the HA coating was resorbed during the implantation period as regards volume and thickness. HA coverage was more reduced on porous-coated than on grit-blasted implants (p = 0.01). No foreign-body reaction or osteolysis was seen. An important finding was that one fifth of the surface with complete resorption of HA coating was replaced by newly formed bone.