Seismic and tracer test data can be combined to estimate the spatial patterns of aquifer properties. We present an algorithm that estimates the geometry of large-scale lithologic zones, the effective hydraulic conductivities and seismic velocities for these zones, and the effective small-scale dispersivity. The heart of this algorithm is our split inversion method, which extracts the geometry of lithologic zones from an estimated seismic velocity field. This method determines the zonation that best matches multiple types of data, such as seismic travel times and tracer concentrations. Although the current implementation of the algorithm uses only cross-well seismic travel times and tracer concentrations, the algorithm could incorporate other data types that are sensitive to changes in large-scale lithology. We demonstrate the approach for two synthetic sandy aquifers, one with interbedded clay lenses and another with interbedded silt and gravel lenses. These examples differ in the uniqueness of the relationship between seismic velocities and hydraulic conductivities. For these examples our algorithm successfully maps interwell heterogeneities and accurately estimates hydraulic and seismic parameters. ¿ American Geophysical Union 1994