Multiple migration pathways and the need to consider several potential siting horizons render the problem of groundwater monitoring network design a difficult task for three-dimensional systems. While the application of quantitative simulation-based approaches for this problem is often impractical due to computational requirements, qualitative approaches can be ineffective because they are highly subjective and typically lack a set of well-defined criteria for locating sampling sites. An analytically based methodology developed in this paper integrates the practical implementation aspects of a qualitative approach with a quantitative analysis in deriving groundwater monitoring networks geared toward early detection of migrating contaminant. Monitoring wells are located among sets of candidate nodes in each of several hydrostratigraphic intervals on the basis of contamination susceptibility. Susceptibility is defined by nodal weights, which are derived on the basis of the locations of sites relative to contaminant source boundaries and potential contaminant plumes. As is illustrated from the results of a case study application, derived monitoring network configurations exhibit two key characteristics: (1) clustering of wells around outlets at contaminant source boundaries and (2) coverage of vacant areas in multiple siting horizons that are susceptible to contamination. ¿ American Geophysical Union 1993