Streamflow simulation provides a powerful tool for developing better information on flow magnitudes and time patterns for the design of water resources projects. A self-calibrating simulation model was applied to about 20 Kentucky watersheds and successfully used to estimate appropriate values for model parameters. The procedure required to develop the necessary input data is explained in sufficient detail to essentially be a users' manual for the program. Physical descriptions of the watersheds are presented, and a number of key watershed characteristics (soil depth, permeability, cover, etc.) were measured. Relationships between model parameters (based on the Stanford Watershed Model) and watershed characteristics were explored. Those which are most promising in estimating parameter values for use in modeling streamflow from ungaged watersheds are presented. Parameter values were also estimated for a series of years spanning a period of intensive urbanization, and statistically significant trends were noted. The results provide a framework for modeling changes in streamflow magnitudes and time patterns associated with urban development. (WRSIC abstract)