A one percent stratified random sample of all Dutch households (comprising 62 000 households) is used to examine mobility in the Dutch housing market. Two techniques, proportional reduction in uncertainty (PRU), and logit analysis are used to evaluate the relative contribution of independent variables in explaining mobility across the tenure types and by housing market regions. The PRU technique is used to select the best variables and to simplify the categories of those variables before logit analyses. Given a reduced set of variables and combinations of categories, logit models are utilized to provide parameter estimates of the contributions of the independent variables. The PRU technique shows that it is possible to make considerable simplifications in the combinations of categories for variables, and the logit models (for owners, public renters, and private renters) indicate distinct differences in the combinations of variables which are most appropriate in each of these sectors. In particular, the logit analysis shows that the relationship between age and mobility is almost linear, but between income and mobility it is curvilinear, and that evaluating interaction effects adds significantly to our understanding of the reason to move or to stay. The models show that it is necessary to treat the three tenure sectors separately, even though age, just as we expected, dominates all sectors in terms of its explanatory power. Although there are not large regional effects, the models change for different regions of the country.