Despite their importance to environmental issues and mineral resources, surficial sediments are largely unmapped at depth in the United States. Full three-dimensional (3D) models of these materials are needed to support hydrologic modeling, geotechnical engineering, and mineral-resource inventory applications. The main source of information for such models in the Midwestern United States is lithology logs from public water-well records. Three-dimensional modeling of lithology was conducted to elucidate the nature and quality of spatial information sourced from water wells. The modeling was conducted on an ∼130-km2 area near Lake Erie in northeast Ohio that contained an end moraine superimposed on a buried glacial valley. An integrated approach to 3D modeling was adopted where traditional interpretive techniques were used to define glacial stratigraphic units (glacial outwash, till, etc.) with geostatistical simulation of lithofacies conducted within the stratigraphic layers. Despite the large amount of variability and noise inherent in water-well data, there were statistical patterns in the lithology records related to glacial stratigraphy. In contrast, the well data provided only minimal information on facies geometry within these units. The spatial structure of facies in the vertical direction was based on thickness statistics from the water-well data and on geologic interpretation for the horizontal direction. Several sequential indicator simulation models were conducted to investigate the effect of grid-cell thickness, stratigraphic trimming, and length-to-thickness ratios on the reproduction of thickness statistics. This case study confirms that water wells are a viable data source for stratigraphic modeling, but better data sources (e.g., outcrops and geophysics) are needed to parameterize facies simulations.