Trends in reported robbery, burglary, larceny, and auto theft rates are modeled, using annual data for the period between 1947 and 1972. A 'criminal opportunity perspective' is used to formulate several substantively meaningful social production functions for these crime rate trends, showing how relatively moderate social changes can generate dramatic increments in the crime rate. For example, the participation of women in the labor force, the incidence of persons living alone, and the presence of lightweight durable goods provide offenders with oppotunities favorable for carrying out illegal acts. Stochastic equations estimating these production functions indicate that the null hypothesis of no autocorrelation of disturbances is consistently accepted. Ex post forecasts of the period from 1973 to 1975 reported crime rates used to gauge the accuracy of the models usually err within a few percentage points. The traditional cross-sectional inverse relationship between socioeconomic variables and crime rates does not necessarily apply to annual changes in the particular crime rates examined here for the post-World War II United States. On the other hand, consideration of changes in criminal opportunity appears useful for specifying equations which perform well statistically, forecast recent crime trends as well as most econometric models, and conform to substantive knowledge about human activity patterns. Furthermore, structural variables producing crime rates interact statistically and demonstrate increasing returns to scale, thus indicating that dramatic increases in crime rates can result from modest changes in trends of predictor variables. When linked to relevant organization goals, forecasts which include criminal opportunity factors can provide technical input emerging from systematic procedures which can aid decisionmakers in the calculation of optimal policy choices. Statistical data, footnotes, and 39 references are included. Sources of data are appended. (Author abstract modified)