It is known that a priori information on the distribution of neutron fluxes over neutron transport models can improve the computational efficiency of Monte Carlo transport calculations via the use of variance reduction techniques. The uncertainty lies in finding a method which obtains this information as fast and effortlessly as possible. A number of global variance reduction techniques were tested on the ITER reference neutronics model for their applicability for performing large three-dimensional transport simulations for ITER. Among these, we propose a new method which is computationally fast and efficient in determining the a priori flux information. The new method, known as MAGIC (Method of Automatic Generation of Importances by Calculation), was found to perform well. The MAGIC method uses an initial analogue run to determine an approximate distribution of particle flux across as much of the problem as possible. The output of this run is then used to produce an MCNP weight window file and used as the input for a subsequent calculation. This iterative stage proceeds until the entire mesh is populated. The MAGIC method was applied to the ITER Tritium Breeding System (TBS) model, which has an extent of over 20 metres, and used to produce a mesh weight window file for MCNP. The weight window was capable of providing high resolution neutron spectra over the entire model domain.