The two groups were not randomly determined, but rather were comparable in demographic characteristics and rates of pre-program arrest. The CONREP group had a significantly lower community-period arrest rate than did the MAXOUT group. Survival rate analyses, using time until arrest and time until revocation as dependent variables, indicate revocations during the first 240 days of community re-entry reduced the potential arrest rate within the CONREP group. Results are discussed in terms of the "criminalization hypothesis" as well as the crime-preventive possibilities of the conditional release mechanism. That revocations were reactive to some of the same kinds of behaviors that resulted in arrest among MAXOUTS is suggested by the "criminalization hypothesis" (Teplin, 1984). For MAXOUT cases to have been involuntarily rehospitalized rather than arrested, they would have to have met the same civil commitment standards as would any private citizen. In California as in many other jurisdictions, these standards have become so stringent and public mental health resources have become so scarce, that mental illness symptoms among the general public have been increasingly dealt with by police arrest rather than by mental health intervention. Conditionally released clients were more likely to have been kept within the mental health treatment system than to have been arrested and subjected to the often inadequate capabilities of most jails to screen and treat mentally ill inmates. 2 figures, 2 tables, and 21 references