A new canine model utilizing an implantable chamber with multiple bone ingrowth channels has been used to study the response of intramedullary bone to various implant materials and surfaces. The first group of dogs received implants containing channels lined by smooth-surfaced coupons of titanium, titanium alloy, sputter-hydroxyapatite-coated (HA-coated) titanium alloy, and polyethylene. A pattern of early initial bone ingrowth by 2 weeks, becoming maximal at 6 to 12 weeks with remodeling to a more mature lamellar bone, and later resorption by 24 weeks was seen for all test groups, with fibrous tissue interfaces covering the smooth test coupons at all time points. Significantly increased bone ingrowth in the sputter-HA coated group was found only at 6 weeks. The second group of dogs received implants with channels lined by surface-roughened coupons of either titanium or plasma-HA-coated titanium, half of which were also packed with a crystalline-HA grouting at the time of surgery. At both 6 and 12 weeks, bone ingrowth was greatly enhanced by the presence of the plasma-HA coating or the crystalline-HA grouting as compared to the uncoated titanium channels. Histologically, bone was seen to bond directly to the plasma-HA coating and the crystalline-HA grouting. A thin fibrous tissue layer was noted between bone and the titanium in most areas, but evidence of direct bone contact to the metal surface was seen. Mechanical testing in tension of intact coupon-bone-coupon units revealed significant strength of the bone-plasma-HA bond, with failure initiating at the metal-HA interface with forces of 15.3 N at 6 weeks, increasing to 44.8 N at 12 weeks. Plasma-HA-lined channels with crystalline-HA packing required similar forces for failure. No significant adhesion strength was noted for the titanium channels at 6 weeks, and only the crystalline-HA-filled channels displayed measurable strength of the bone-titanium interface at 12 weeks, with a force of 9 N needed for failure.