Current upgridding techniques usually require the scale-up factor to be known a priori and, in practice, the scale-up factor is selected rather subjectively. The main consideration is that the scale-up factor should be large enough for practical use and not so large that the results are likely to be very inaccurate. A more objective approach to selecting the scale-up factor is desirable. This paper proposes a novel approach to determine a more appropriate scale-up factor for a coarse model. The new approach is based on an analysis of the flows in the fine-scale model. The regions that make a significant contribution to the flow are identified using a thresholding technique. A range of scale-up factors is tested and high flow regions are identified in each case. The percentage overlap (denoted by Ro) of the high flow regions in the coarse and fine grids is then calculated. The value of Ro can be used as a predictor of the accuracy of upscaling and the appropriate coarse grid size can thus be evaluated. When the upscaling is performed using a global method, there is little extra computational cost. This method is suitable for use in multi-phase flow simulations of 3D heterogeneous reservoir models. The technique is demonstrated using a simulation of a waterflood in a highly heterogeneous 2D model.