In this study, we consider a dynamic demand model for international tourism. The model is used to estimate separate demand functions for hotel and cottage visitors from different countries, with a multivariate structural time series model. Among other things, the estimated models are used to simulate the effects on international tourism demand in Sweden of an increase in the value added tax on typical tourism products. The results reveal that the price sensitivity differs considerably between various visitor groups. The largest effect is found for Norwegian visitors, while there is no significant effect for Danish ones. The forecast accuracy of the demand model is also evaluated. The results indicate that a pure structural time series model performs as well as a model with explanatory variables. © 2004 Taylor & Francis Group, LLC.