The ability of the microbond technique to characterise changes in the physico-chemical structure of the interface between fibre and matrix has been checked using eight epoxyde/glass fibre systems differing by their matrix chemistry, fibre surface treatments, and fibre diameter. It has been shown that the widely used average IFSS method can lead to biased results. The test is now considered as giving mode I or mixed-mode properties of the interface and not only a mode II interfacial toughness or interfacial shear strength (IFSS). Energy approaches are thus be preferred to stress criterion models. The suitability of six theoretical models was checked. Difficulties were found in determining a parameter or method effectively representative of the physico-chemical structure of the interface. The model providing the most reliable results was that of Scheer and Nairn. Significant plastic flow of the polymeric droplet was observed, leading to a questioning of the hypotheses of ideal elastic components. (C) 2001 Elsevier Science Ltd. All rights reserved.