The mechanical behaviour of peripheral nerves is known to vary between different nerves and nerve regions. As the field of nerve tissue engineering advances, it is vital that we understand the range of mechanical regimes future nerve implants must match to prevent failure. Data on the mechanical behaviour of human peripheral nerves are difficult to obtain due to the need to conduct mechanical testing shortly after removal from the body. In this work, we adapt a 3D multiscale biomechanical model, developed using asymptotic homogenisation, to mimic the micro- and macroscale structure of a peripheral nerve. This model is then parameterised using experimental data from rat peripheral nerves and used to investigate the effect of varying the collagen content, the fibril radius and number density, and the macroscale cross-sectional geometry of the peripheral nerve on the effective axial Young's moduli of the whole nerve. Our results indicate that the total amount of collagen within a cross section has a greater effect on the axial Young's moduli compared to other measures of structure. This suggests that the amount of collagen in a cross section of a peripheral nerve, which can be measured through histological and imaging techniques, is one of the key metrics that should be recorded in the future experimental studies on the biomechanical properties of peripheral nerves.