This paper tests the deterrent ( or displacement) effects of preventive patrol upon criminal and non-criminal disorderly behaviors at high-crime locations ("hot spots") using observational data collected during a preventive patrol experiment in Minneapolis from December 1988 to November 1989. The analyses reveal that the immediate presence of uniformed police directly reduces the outbreak of disorderly conduct at hot spots, but this effect is contingent upon raising the overall level of proactive presences at hot spots. Increasing patrol levels at hot spots also produces residual deterrence which decreases disorder during times when police are not present at these locations. Such residual decreases in disorder are larger than the direct deterrent effects of police presence when patrol is at normal levels. Further, direct and residual deterrence generated by patrol are stronger for criminal acts than for a combined measure of criminal and non-criminal disorderly behaviors. The analyses employed survival models to estimate the effects of specific instances of patrol presence upon the time to the first disorder (criminal or non-criminal) after police depart from a hot spot. Using presences up to 20 minutes in length, these models reveal that longer presences increase survival time, thus enhancing residual deterrence. However, there is evidence this effect decreases after presences pass about 14 minutes in duration. Moreover, stops must be about 10 minutes in length in order to produce significantly better survival times than those produced by driving through a hot spot. The theoretical and policy implications of these results are discussed.