This paper describes a procedure for calculating the matrix-fracture transfer shape factor for dual porosity models for fractured reservoirs. The procedure is implemented in 3D and uses the Continuous Time Random Walk (CTRW) method. In previous works the method has been implemented only in 2D and has required a mesh produced by triangulation of the 2D fracture network. The CTRW method used here is extended to 3D in a simpler implementation with rectangular grid cells, which significantly reduces computation effort, but otherwise relies on the same principles. The method is based on modelling of random walk of particles through the matrix media. For each particle the random walk is terminated when the particle encounters a fracture and the random walk time is recorded. The average random walk time for many particles is used for calculating the shape factor for the investigated volume. The calculated shape factor with this new method is compared with analytical estimates for idealized regular networks of fractures, and with flow simulation-derived shape factors for more realistic fracture networks. The results demonstrate that the new method reproduces analytical and numerical results in simple fracture models from other studies and, furthermore, is capable of analysing very complex 3D fracture models and their properties. The method has potential in traditional dual porosity simulation in oil and gas studies, and also in transport modelling where diffusion processes are the driving force in exchange between matrix and fractures. The method also has potential as a general characterization tool in fractured rock systems, where the shape factor and its distribution can be used as a qualitative descriptor of the rock matrix block-geometry and as an indicator of the behaviour of matrix-fracture exchange processes.