In recent years techniques have been developed to obtain integrated water vapor along slant paths between ground-based Global Positioning System (GPS) receivers and the GPS satellites. Results are presented of an observing system simulation (OSS) to determine whether three-dimensional water vapor fields could be recovered from a high-resolution network (e.g., with 40-km spacing) of GPS receivers, in combination with surface moisture observations and a limited number of moisture soundings. The paper describes a three-dimensional variational analysis (3DVAR) that recovers the moisture field from the slant integrated water vapor and other observations. Comparisons between ``nature'' moisture fields taken from mesoscale models and fields recovered using 3DVAR are presented. It is concluded that a high-resolution network of GPS receivers may allow diagnosis of three-dimensional water vapor, with applications for both positioning and mesoscale weather prediction.