The present study makes part of a research project on the geological and geomechanical characterisation of weak cataclastic fault rocks encountered in underground excavation sites in the Alps. The project is composed of a geological and a geomechanical part, split up in two PhD theses and realised in parallel at the Laboratory of Geology (GEOLEP) and the Laboratory of Rock Mechanics (LMR) at the Swiss Federal Institute of Technology in Lausanne. By their very contrasting and unfavourable mechanical behaviour in underground excavations, weak cataclastic fault zones influence directly the excavation method, the safety of the working site, the choice of tunnel support, the long-term behaviour and the costs of the infrastructure. The construction of galleries across cataclastic fault zones bears furthermore the potential risk to modify the local and regional hydrogeological regime. Weak cataclastic fault rocks show a large scatter of geological and geomechanical properties. Due to their low cohesion, they alter rapidly when exposed to weathering. They can hence be studied and sampled in natural outcrops and exploratory boreholes only with great difficulty. Analyses and laboratory tests are laborious and require special sample preparation techniques and equipments. For these reasons still little is known about their geological, hydrogeological and geomechanical behaviour. The purpose of this study is to develop a detailed and objective differentiation and characterisation method for weak cataclastic fault rocks. Classifications and terminologies of (weak) cataclastic fault rocks are in fact still controversial. After a review of fault rock definitions and rock deformation mechanisms, the term kakirite has been retained in agreement with Heitzmann (1985) and Wyder (1997) as a general designation for weak, cohesionless cataclastic fault rocks. This term has been approved recently by the Swiss Society of Engineers and Architects (SIA) to designate fault zones made up of "loose, structureless material" (SIA Norm 199, 1998). Cataclastic fault zones have been studied and sampled in galleries currently under construction in different geological and tectonic contexts in the Alps. Research has been concentrated mainly on fault zones occurring in quartzo-phyllitic host rocks. For comparison, fault rocks from greenstones and serpentinite have been analysed too. To improve the understanding of the geomechanical and hydrogeological behaviour of weak cataclastic fault rocks, a new quantitative characterisation method has been developed. In a simplified approach, the geomechanical behaviour of (fault) rocks is considered to be controlled by two factors, namely the: mineralogical composition (mwVh), and rock fabric (TC, MC) The proposed mineralogical and structural characterisation of cataclastic fault rocks is referring to the meso- to microscopic scale, respectively to a scale comparable with common geomechanical laboratory tests. Mineralogy is expressed quantitatively by means of the mean weighted Vickers hardness, mwVh (Calembert et al., 1980a). The relative proportions of the principal rock forming minerals have been determined by semi-quantitative XRD-analyses. The mineralogical composition of kakirites, adjacent host rock samples and sieved kakirite grainsize fractions has been determined. Comparing the composition of the different grainsize fractions a clear dependency of grainsize on mineralogical composition has been found in kakirites. Quartz content is in general considerably higher in the clast than in the matrix fraction. The later is in fact composed mainly of phyllosilicates. This mineralogical grainsize fractionation is interpreted to be controlled by the original host rock fabric, the strength contrast between the involved mineral phases and alteration processes. Except kaolinite (Varzo), no other alteration mineral phases have been identified. The presence of (swelling) clay-minerals in very low proportions which cannot be detected by the performed whole rock XRD-analyses cannot be excluded however. Kakirite samples sieved for XRD-analyses show an extended grainsize distributions with in general high proportions of silt/clay grainsize fraction (mean 45%). This may explain the often observed very low permeability of kakirite zones. Their hydrogeological role is illustrated in the last part of this work by the means of a case history of a fault zone crossed by the gallery of Cleuson-Dixence. It demonstrates the often observed groundwater flow barrier-effect of cataclastic fault zones and its impact on the excavation work and the local hydrogeological regime. Rock fabric of kakirites has been characterised by thin section analyses with subsequent semiautomatic image analysis. It is based on a two-phase "clast-matrix" model, considering kakirites to be composed of varying proportions of "hard" clasts and "weak", fine-grained matrix. Clast properties (density, shape and relative orientation) are quantified by the texture coefficient (TC) of Howarth & Rowlands (1987). Pre-existing discontinuities in the matrix are characterised by a newly developed matrix coefficient (MC), involving the discontinuity density, roughness, mean length and relative orientation. The combined structural approach involving the MC and TC has shown to be appropriate to characterise and differentiate a wide range of weak cataclastic fault rocks by their microscopic rock fabric. Two opposite kakirite fabric types can be distinguished: first, a granular fabric, characterised by clasts taken in a fine-grained, isotropic matrix. Second, a discontinuity dominant, often schistose fabric wherein no clasts can be identified. A gradual transition is observed in-between these two end-member types. In order to assess the validity of the developed characterisation method, a first attempt of correlation has been made between the mineralogical-structural parameters and corresponding rock strength of kakirites tested by Habimana (1999). The found correlation is a strong motivation to continue the started research on cataclastic fault rocks with the presented mineralogical and structural approach.