OBJECTIVES: Vaccines play an important role in the induction of herd protection. To date, few studies address this topic with reliable evidence. A hypothesis-generating modelling study of PCV-vaccine was carried out using selected variables thought to impact on herd effect. METHODS: A simple dynamic model was developed to understand potentially important factors impacting herd effect. Two patient age groups (0-4, 5 +) and 2 types of infection (i.e. vaccine and non-vaccine serotype) were considered. Parameters such as relative force of infection (FOI) of serotypes, vaccine coverage (i.e. serotype distribution and population coverage of the vaccine), transmission rate between age-groups and co-colonisation rate were evaluated. Key assumptions relate to constant birth/ death rates, serotype transmissibility, vaccine-induced protection duration (10y), similar transmission pattern and average carriage duration between vaccine serotypes and non-vaccine serotypes. RESULTS: Net herd effect is predicted to vary between 8% and 72% with different assumptions on cocolonization (factor between 0.1 and 0.7) and relative FOI for non-vaccine type (decreased by 1% to 10%) with a vaccine coverage of 70%. Simple intuition would suggest that greater vaccine coverage is associated with greater herd effect in the long term. However, our model suggests this occurs only when the cocolonization of vaccine and non-vaccine serotypes represent more than 8% of the S. pneumoniae carriers. Therefore, the converse is actually true, i.e. more vaccine coverage is associated with less herd effect when cocolonization is less frequent (i.e. below 7-8% as the literature would suggest). Demographics, contact patterns between individuals and the relative FOI of non vaccine serotype to vaccine serotype also heavily determine that effect. CONCLUSIONS: All parameters tested in our model impact strongly on the predicted net herd effect. Vaccine characteristics such as coverage are not the only and most influential factor that affects herd protection in a given population.