A theory of the ways in which changes in criminal opportunity affect these property crime rates is developed. Definitions and postulates are presented and used to derive a main theorem indicating that, other things being equal, a decrease in the density of the population in physical locations which are normally the sites of primary groups' routine activities should produce an increase in criminal opportunities and hence in property crime rates. Corollaries to the main theorem are presented and tested using relevant independent and control variables, such as the residential population density ratio, the unemployment rate, age structure, total consumer expenditures, and automobiles per capita. Stochastic difference equations are used to evaluate the theory and indicate that the models implied by the theory exhibit good statistical fit to the recorded property crime rates at issue over the 26-year estimation period (1947-72). In addition, these models provide reasonably accurate ex post forecasts of observed annual property crime rates over the 5-year forecast period, 1973 through 1977. Forecasts based on the models indicate that the robbery and automobile theft rates should drop substantially in the 1980's from their recent levels, whereas the burglary rate may continue to grow or at least drop less. Tabular and graphic data and 51 references are provided. (Author abstract modified)