Fibre reinforced composites have gained increasing technological importance due to their versatility, which lends them to a wide range of applications. These composites are useful because they include a reinforcing phase in which high tensile strengths can be reached, and a matrix that allows one to hold the reinforcement and to transfer the applied stress to it. It is a well-known fact that such materials can have excellent mechanical, thermal, and electrical properties that make them widely used in industry. During the manufacturing process, fibres adopt a preferential orientation that can vary significantly across the geometry. Once the suspension is cooled or cured to make a solid composite, the fibre orientation becomes a key feature of the final product since it affects the elastic modulus, the thermal and electrical conductivities, and the strength of the composite material (Pipes et al., 1982;Agari et al., 1991).With the increase in composite materials usage, interest in the rheology and processing of fibre suspensions has increased significantly. This growing interest stems from the need to model the flow of fibres as accurately as possible in order to design and control manufacturing processes that generate favourable fibre orientations states, which will ultimately lead to the best mechanical and thermal properties of the composite. The rheological behaviour of a suspension of rigid particles has been the subject of a considerable amount of research (Stover et al., 1992;Becraft and Metzner, 1992;Bay and Tucker, 1992;Ramazani et al., 1997). Most of the existing theoretical work is based on the early studies of Jeffery (1923) and Batchelor (1971).Continuum models are often used to describe the orientation of fib r e s in a flo w. These models are hyperbolic in nature, and often lead to sharp b o u n d a ry layers and singularities even for simple flow geometries. In order to avoid this limitation, many authors adopted a simplifying approach known as the fibre-aligned assumption to study fibre suspension flows in various contraction and expansion geometries (see for example Lipscomb et al., 1988;Chiba et al., 1990; Baloch and We b s t e r, 1995). In this approach, it is assumed that the fibre is completely aligned with the streamlines, and therefore one has to solve only for the velocity field from which the fibre orientation can be deduced. The same approach has also been used by Rallison and Keiller (1993) to investigate