This study aims at studying the effect of surface roughness on bone attachment of coin-shaped titanium implants. All implants in this study were blasted with TiO(2) particles of 180-220 microm, and then divided into three groups. One group had no further surface treatment whereas the other two groups were subsequently etched with hot hydrochloric acid (0.01M or 1M). The surface topography of the implant specimens was examined by SEM and by a confocal laser scanner for a numeric evaluation of S(a), S(t) and S(dr). The ranging implants in the three groups differed significantly in surface structure. The implants with modified surfaces were then placed into the tibias of 12 rabbits (n=16). After 8 weeks healing, the attachment of bone to implants were examined using a standardised tensile test analysis. The implants that were only blasted (positive control) showed significantly better functional attachment (p<0.001) than the acid etched. Implant surfaces etched with 1M HCl solution had the lowest retention in bone. There was a negative correlation between the increasing roughness and mechanical retention in bone of the implants in this study. The results support observations from earlier studies that suggested an optimal surface roughness for bone attachment, identified by in situ tensile tests and expressed as the arithmetic mean deviation (S(a)), to be in the range between 3.62 and 3.90 microm and that a further attachment depended on mechanical interlocking between bone and implant.