After the recent introduction of preventive detention in Germany, the Swiss national council has passed a similar law. In both these countries, as well as other nations, there is a controversial debate on post-sentence preventive detention of offenders who are found to be very dangerous during enforcement of sentence. Empirical results as to if and how far post-sentence preventive detention is sensible and appropriate are lacking, due to the scarcity of such cases and the need for systematic observation and long survey periods. In all cases used in the present study, a very high dangerousness was assessed during enforcement of sentence, but release had to be granted for legal reasons. With a total of nine cases (all registered from 1997 to 2005), these high-risk offenders represented a small and narrowly defined group, ie 2% of all violent and sex offenders administered by the Canton of Zurich. They also differ from the main population regarding socio-demographics and psychiatric and criminological aspects. It was possible to evaluate the progress of eight of the released offenders in a follow-up study. All eight offenders re-offended with grave violent and sex offences, seven within a year of their release, causing a total of twenty-four victims. The present study supports the supposition that at least some categories of very dangerous offenders can be reliably detected. It also supports the notion that systematic risk assessment during enforcement of sentence, using all obtainable information, can help prevent serious violent and sex offences. Furthermore, the results also illustrate the necessity of post-sentence preventive detention of a small number of high-risk offenders as an important measure for the protection of potential victims.