In the manufacture of high performance carbon fibre re-inforced plastics (CFRP) good adhesion between fibre and matrix is essential in order to ensure optimum stress transfer across the resulting interfaces. The overall interfacial interaction is due to specific (chemical) and non-specific Van de Waals interactions between fibre surface atoms and matrix materials. Fibre surface topography, roughness and porosity, are also likely to contribute to interracial bonding by the mechanical keying together of the two components. In order to enhance the surface reactivity of carbon fibres the concentration of oxygen functionality is often increased by an oxidation step during manufacture. The process may also remove loosely bound, low molecular weight, carbon species from the fibl-e surfaces leaving a mechanically more suitable surface for adhesion. Detailed studies of the factors which determine interracial interactions are essential to the understanding of mechanical properties and modes of failure of composiite materials. We are engaged in a long-term investigation of how the surface chemistry of carbon fibre, and other fibre and particulate reinforcing materials, influences interface formation and the resulting mechanical properties of composites. In this communication we discuss the changes in surface chemistry, brought about by commercial anodic oxidation treatment, of PAN fibres and their effect on apparent interlaminar shear strength (1LSS) data for an epoxy composite. Batches of Akzo HTA polyacrylonitrile (PAN)based carbon fibres have been studied. Surface chemistry and contact angle data are compared for 0% and 100% (HTA 500X and 5001, Lots 11019 and 11011 respectively). Unoxidized fibres have quoted tensile strengths of 4012 (+1.5)MPa and tensile moduli of 239 (_+1.5) GPa. Oxidized materials have tensile strength of 3900 (+124)GPa and tensile moduli of 245.7 (+4.0)GPa. Unidirectional fibre test pieces have been produced from both fibre types after application of a bis-phenyl A size (HTA 5130 and 5131 same strength and modulus as above). A standard toughened epoxy (Ciba-Geigy F914) has been used as matrix. Apparent ILSS data have been calculated using the three-point bend test, short beam, method according to ASTM D2344-84. X-ray photoelectron spectroscopy (XPS) measurements have been carried out on a VG MK 1 Escalab using A1 K1 X-rays (energy 1486.6 eV) at a power of 200W in a residual vacuum of 1.333 × 10 -6 Pa. Binding energy resolution has been increased by removal of the X-ray line shape using in-house software on an IBM-AT computer, for functional group identification. Full experimental details have been published elsewhere [1]. Oxidation leads to a factor of 2 increase in measured surface oxygen (Table I). High energy resolution carbon ls spectra (not shown) show this to be distributed as hydroxyl/ether (C-O), carbonyl (C~---O) and carboxylic acid/ester(COOH/COOR). As expected both fibre types have very similar surface compositions after application of size (Table I). However, the oxygen levels recorded are slightly higher than those expected from a stoichiometric bisphenyl A layer (-11%). The 0% oxidized fibres generally have surface nitrogen levels of 2-3 at % which, after this method of oxidation, does not change significantly. In previous studies we have found that anodic oxidation in ammonium bicarbonate solution gives a similar increase in surface