The aim of this study has been to develop processing techniques for novel bioresorbable thermoplastic cellular composites, reinforced with continuous inorganic fibres, for use as biodegradable scaffolds for bone tissue engineering. As the implanted composite scaffold starts to degrade, it should be progressively replaced by natural bone tissue, which should consequently support increasing proportions of the applied mechanical loads. In order to simulate natural high performance porous structures such as bone or light wood, composites with simultaneous gradients in fibre content and porosity were studied. In order to carry out investigations over wide ranges of fibre content and porosity and to ensure biocompatibility and bioresorbability, it was decided to focus on foamable poly(L-lactic acid) (PLA) and continuous glass fibres. As well as improved mechanical reinforcement due to the high aspect ratio, the use of continuous glass fibres offers the possibility of achieving higher loadings than with short fibres or particles, increasing the potential range of mechanical properties. For processing the composites, a fibre winding technique developed in house was used to prepare unidirectional preforms with gradient structures tailored at the fibre bundle level. PLA fibre bundles, containing 36 monofilaments were prepared on a spinning line, and mingled with glass fibre bundles comprising 250 or 1000 monofilaments. During the winding process, the PLA fibres were intimately mixed with the reinforcing fibres, to ensure a good impregnation during subsequent processing steps. Moreover it was possible to maintain the relative positions of the fibres during pre-consolidation of graded composites. The resulting preforms had a porosity of 25% and could be further consolidated to decrease the porosity, or foamed in an autoclave to obtain cellular composites (foams) with porosities of up to 90%. It was observed that, independently of the applied foaming conditions, the presence of continuous fibres reduced the porosity by an amount equivalent to twice the fibre volume fraction. This processing method provided composites with a simultaneous gradient in fibre content and porosity. In order to predict the Young's modulus and the collapse strength of the cellular composites a model was proposed, which combines the buckling of composite columns within a porous structure with the property predictions for neat polymer foams. In neat foams with a porosity of 60%, the Young's modulus was calculated to be 0.25 GPa. By adding a fibre volume fraction of 14% the modulus reached 1.2 GPa. The respective collapse strengths were 5 MPa and 16 MPa. Experimental results were consistent with the predictions and validated the model. The model determines the fibre content required for composites with a given porosity and stiffness. Gradient cellular composites were obtained with a range of porosity between 50% to 90% and a range of stiffness between 100 and 1500 MPa. Furthermore the processing parameters can be adjusted to maintain a desired porosity when fibres are added. Thus, composites that fulfill the mechanical requirements for bone scaffolds can be achieved in a rational manner. The long term behaviour of the foams was tested under physiological conditions. Viscoelasticity tests provided evidence for the important role of the unidirectionally oriented fibres. Whereas the cells in neat foams collapsed after at most two days when loaded at 0.69 MPa, the fibre reinforced foams resisted longer and their creep strain was well described by a power law. The fatigue properties were compared to reported values of natural trabecular bone. None of the neat foams with a porosity of 74% resisted to the first load cycle of 7.3 MPa. With a fibre volume fraction of 11% and a comparable porosity of 72%, the composite scaffolds resisted for at least 11'000 loading cycles. To compare, trabecular human bone failed after 1909 cycles, when loaded with 8.5 MPa. It was concluded that the reinforced composites have the potential to replace natural bone. Preliminary in vitro results with human foetal osteoblasts and in vivo tests on femurs of rats were promising because the cells proliferated well on the composite substrates. Further investigations for applications in bone tissue engineering can be planned. Furthermore, the studied process and the obtained results are envisaged for other material systems to prepare cellular composites for novel lightweight applications in transportation and building industry.