The goal of this study is to develop a new methodology that enables decision makers to visualize and quantify the tradeoff between cost and uncertainty for sites undergoing long-term monitoring. Monte Carlo simulation is used to predict discrete cumulative probability distributions (cdf's) for the dissolved contaminant concentrations at every available monitoring location at a site. Indicator kriging is then used to evaluate the local uncertainty associated with sampling subsets of the available monitoring locations. A Non-dominated Sorted Genetic Algorithm (NSGA) searches for sampling plans that are non-dominated in terms of the two objectives: (1) minimizing sampling costs and (2) minimizing relative local uncertainty. The NSGA evolves the Pareto optimal frontier that represents the optimal tradeoff between sampling costs and relative local uncertainty. Each point on the Pareto front can be decomposed into a sampling design's cost and a spatial mapping of its local uncertainty. This methodology will be applied to the Williams Air Force Base in Arizona. Copyright ASCE 2004.