We used a simulation model to study the spatio-temporal dynamics of a potential rabies outbreak in an immunized fox population after the termination of a long-term, large-scale vaccination program with two campaigns per year one in spring and one in autumn. The 'worst-case' scenario of rabies resurgence occurs if rabies has persisted at a low prevalence despite control and has remained undetected by a customary surveillance program or if infected individuals invade to the control area. Even if the termination of a vaccination program entails such a risk of a subsequent new outbreak, prolonged vaccination of a wild host population is expensive and the declining cost-benefit ratio over time eventually makes it uneconomic. Based on the knowledge of the spatio-temporal dynamics of a potential new outbreak gained from our modelling study, we suggest "terminating but observing" to be an appropriate strategy. Simulating the decline of population immunity without revaccination, we found that a new outbreak of rabies should be detected by customary surveillance programs within two years after the termination of the control. The time until detection does not depend on whether vaccination was terminated within the fourth, fifth or sixth years of repeated biannual campaigns. But it is faster if the program was completed with an autumn campaign (because next-year dispersal then occurs after a noticeable decrease in population immunity). Finally, if a rabid fox is detected after terminating vaccination, we determine a rule for defining a circular hazard area based on the simulated spatial spread of rabies. The radius of this area should be increased with the time since the last vaccination campaign. The trade-off between the number of foxes potentially missed by the emergency treatment and the cost for the emergency measures in an enlarged hazard area was found.