OBJECTIVES: In reimbursement files models estimate the events avoided over time with new interventions. Simulations are performed in the absence of long-term observational data. In this study observed hospitalisation data prior- and post-rotavirus vaccination in Belgium were compared to predicted (i.e. modeled) results after 1,2 and 3 years. METHODS: A Markov cohort model estimated over time the number of hospitalisations potentially avoided with rotavirus vaccination in Belgium. We adjusted the model to the observed vaccine coverage and to RotarixTM vaccine efficacy from clinical trials. The obtained modeling results were compared with observed data collected from 9 Belgian hospitals (2 years pre- and 3 consecutive years post-vaccination). The outcomes of both studies are expressed as a percentual decrease in hospitalisations for 2 age-groups (<2 and <5 years old), after each year post-vaccination. The differences are presented in absolute and relative (%) differences for each subsequent year post-vaccine launch. RESULTS: The observed data provided slightly better outcomes than the modeled results. After 1 year the absolute difference in decrease of hospitalisations between observed and modeled data was 3% (11%) for both age-groups. After 2 years the absolute differences were 3.5% (6%) and 6% (12%) for the 2y and 5y age-group respectively. In the last observation year the absolute differences were respectively 2.8% (4%) and 9.3% (13%). CONCLUSIONS: Compared with the observed data the model estimates are conservative. The more favorable observational results are explained by the indirect vaccine effect on non-vaccinated age-groups which is not captured by the static model. The relatively better model fit in the <2y old with increasing time is explained by the accumulated vaccine impact over time: from 26% (1st year) to 68% (3rd year) . Rotarix is a trademark of the GlaxoSmithKline group of companies.