The objective of the research presented here is to determine how the representation of the spatial variability of aquifer properties affects ground water flow and contaminant transport modeling at the United Creosoting Company site in Conroe, Texas. Simulations are performed using spatial distributions of aquifer properties generated by each of three methods: constant value estimation, polynomial trend surface estimation, and kriging point estimation. The significance of using a representation of an aquifer property is dependent on the modeling purpose. In natural gradient ground water flow simulations the representation of hydraulic conductivity and bottom elevation significantly affected model calibration. The representation of aquifer properties generally did not affect model calibration of natural gradient contaminant transport modeling, but do affect the shape and size of the simulated plume. The representation of aquifer properties was very useful for identifying feasible ground water recovery options.