OBJECTIVES: Herd immunity describes a phenomenon in the area of communicable diseases. Pathogens are spread by infected persons. Protecting a part of the population - for example via vaccination - lowers the overall appearance of pathogens as these people cannot spread pathogens any more. Not protected people profit by fewer contacts with pathogens and further a lower number of infections for them can be expected. Classic Markov model cannot provide herd immunity as a result. In this work we propose calculations of herd immunity, create a model that is able to simulate epidemics and show herd immunity dynamically in different states of the model. Appearance of herd immunity is very disputed because it (1) cannot be measured directly in real life and (2) depends on several factors. METHODS: Classic Markov models require herd immunity as a static input parameter that cannot be provided. The developed agent based model includes single persons with different infection states and a single pathogen. Every agent is part of a social contact model. It is possible to simulate scenarios without vaccinations and with different vaccination strategies. Herd immunity as a result of the dynamic model is calculated as the reduction of the carrier rate of non-vaccinated persons for a certain vaccination strategy compared with the scenario without vaccinations. RESULTS: Results show herd immunity as simulation result depending not only on vaccination strategies but also on other system parameters. Further work extends the social contact structure with places like households, schools or workplaces that are expected to have an impact on herd immunity as well. CONCLUSIONS: Results can be implemented in systems for calculating new strategies for vaccination programs. Current work considers two or more concurrent serotypes where herd immunity and serotype replacement affects each other. In this case different definitions of herd immunity are possible.