AIM: To investigate bone-to-implant bonding for some novel surface modifications with a hierarchic structure and to correlate the in vivo results with surface roughness parameters. MATERIALS AND METHODS: Newly developed implants surfaces were tested in rabbits and compared with the commercially available OsseoSpeed™ (OS) implant. The blasted test samples were subjected to treatment in oxalic acid (AT-II), followed by subsequent etching in hydrofluoric acid (AT-I). Scanning electron microscopy and X-ray photoelectron spectroscopy were used to characterize the surface topography and chemical composition of the implants. Biomechanical testing after 6 weeks of healing was complemented with the quantification of fluorochromes and the results were subjected to a multivariate statistical analysis. RESULTS: The results show, both with biomechanical- and with histomorphometrical tests, that the AT-I implants with different surface roughness at the micro (blasting), submicro (shallow cavities) and nanolevels (precipitates) have a greater bone tissue integration compared with the AT-II- and OS implants. The 2D bone-to-implant contact (BIC) data were in accordance with the 3D removal torque (RTQ) results even if the former were deduced from implants located in spongeous-type bone and the latter in cortical bone. The increase in RTQ values for the test samples AT-I and AT-II compared with the reference complies with the slightly higher S(a) values for these surfaces. CONCLUSIONS: Using a combination of conventional methods with novel quantification of florochrome and multivariate analysis, the influence of surface roughness on different levels could be discriminated. The RTQ and BIC values show that the most hierarchical structure with submicro cavities and nanoscale precipitates possesses the most favourable osseointegration properties.