Unfortunately, they provide little guidance as to the amount of time that is required for these effects to be realized. In this paper, autoregressive integrated moving average (ARIMA) time-series modeling techniques, which necessitate making minimal assumptions concerning the lag structure one expects to find, are used to examine crime-arrest relationships in 1967-1980 aggregate data for robbery, burglary, grand larceny, and auto theft in Tulsa and Oklahoma City, Oklahoma. Consistent with deterrence and incapacitation theses, robbery arrests had a negative effect on robbery offenses in both areas; but no other significant relationships between crimes and arrests were found. These results add to a growing body of research that raises serious doubts about the efficacy of cross-sectional studies of the crime-arrest relationship. 3 tables and 25 references. (Author abstract modified)