Regression trees are tree-based models used to solve those prediction problems in which the response variable is numeric. They differ from the better-known classification or decision trees only in that they have a numeric value rather than a class label associated with the leaves. Model trees are an extension of regression trees in the sense that they associate leaves with multivariate linear models. In this paper a method for the data-driven construction of model trees is presented, namely the Stepwise Model Tree Induction (SMOTI) method. Its main characteristic is the induction of trees with two types of nodes: regression nodes, which perform only straight-line regression, and splitting nodes, which partition the sample space. In this way, the multivariate linear model associated to each leaf is efficiently built stepwise. SMOTI has been evaluated in an empirical study and compared to other model tree induction systems.