Background. Wear particles are considered to be the major culprit for the aseptic loosening. Their characterization is thus crucial for the understanding of their bioreactivity and contribution to the development of aseptic loosening. Methods. Metal wear debris particles were analyzed directly in periprosthetic tissue resins by scanning electron microscopy (SEM) combined with back-scattered electron imaging (BSE) and energy dispersive X-ray spectroscopy (EDS). Four groups of tissue samples retrieved at revision operations of loosened hip implants with different bearing surfaces (metal-on-metal, ceramic-on-polyethylene and metal-on-polyethylene), and different material of the femoral stem (Ti alloy, CoCrMo and polymer combined with stainless steel) were investigated. Tissue samples were first analyzed histologicaly. Sections from the same paraffin blocks were then carbon coated and analyzed using SEM/BSE/EDS method. Results. Metal particles were detected in all samples. Their composition corresponded to the composition of the implant components. The gradation of metal particles ranged from +1 to +3. A considerable number of big metal particles were actually agglomerates of submicron particles visible only at higher magnification. The clustering of particles was observed primarily for CoCrMo and, to a lesser extent, for stainless steels particles. The median sizes of CoCrMo clusters in two groups of samples were 2.9 1.8 m (range, 0.5 to 7.6 m) and 3.2 1.0 m (range, 1.9 to 5.4 m). The effect of clustering was not observed for Ti particles. The median sizes of individual Ti particles determined in two groups of samples were 2.5 3.6 m (range, 0.4 to 17.3 m) and 4.3 2.8 m (range, 0.8 to 11.0 m). Conclusion. Scanning electron microscopy combined with back-scattered electron imaging is an appropriate and selective method to recognize metal particles in tissue sections, without being destructive to specimens. When the size of the particles is considered, however, it should be differed between the size of individual particles and size of clusters of particles. Besides its benefits, this study has some limitations: the detection of particles smaller than 0.4 m is difficult, and this method cannot be used to identify polyethylene particles