The objective of this research project has been to investigate the potential for using temperature-dependent (thermal) precipitation of chemicals to reduce the porosity and? permeability of porous rocks. The method consists of injecting hot water that is saturated in a chemical that will precipitate upon cooling. Through this process, the permeability of thief zones in oil reservoirs could be reduced, allowing improved recovery by secondary and tertiary recovery processes. The chemical literature was reviewed for environmentally safe chemicals that have a suitable temperature-dependent solubility for the thermal precipitation process. Four suitable chemicals were identified: boron oxide, potassium carbonate, sodium borate, and potassium chloride. An experimental apparatus was constructed to test the thermal precipitation process at high temperatures and pressures. Data was collected with clastic Berea sandstone cores using two chemicals: potassium carbonate and sodium borate. Data was also collected with limestone cores using potassium carbonate. The porosities and permeability's were measured before and after being treated by the thermal precipitation process. A theoretical study of the process was also conducted. A model for predicting the fractional reduction in porosity was developed that is based on the temperature-dependent solubility of the chemical used. An empirical model that predicts the fractional reduction in permeability in terms of the fractional reduction in porosity was then developed for Berea sandstone. Existing theoretical models for estimating the permeability of porous media were tested against the measured data. The existing models, including the widely-used Carman-Kozeny equation, under predicted the reduction in permeability for the thermal precipitation process. This study has shown that the thermal precipitation process has considerable potential for the controlled reduction in porosity and permeability in geologic formations. A design study to determine how the process would work in the field is recommended.