We describe a fast method for estimating flow through a porous medium with a heterogeneous permeability distribution. The main application is to contaminant transport in aquifers and recovery of oil by waterflooding, where such geological heterogeneities can result in regions of bypassed contaminants or oil. The extent of this bypassing is normally assessed by a numerical flow simulation that can take many hours of computer time. Ideally the impact of uncertainty in the geological description is then evaluated by the performing of many such simulations using different realizations of the permeability distribution. Obviously, a proper Monte Carlo evaluation may be impossible when the flow simulations are so computationally intensive. Consequently, methods from statistical mechanics, such as percolation theory and random walkers (such as diffusion-limited aggregation), have been proposed; however, these methods are limited to geological heterogeneities where the correlation lengths are smaller than the system size or to continuous permeability distributions. Here we describe a growth model that can be used to estimate the breakthrough time of the water (and hence the sweep efficiency) in most types of geologically heterogeneous rocks. We show how the model gives good estimates of the breakthrough time of water at the production well in a fraction of the time needed to perform a full flow simulation.