Regression analysis is a frequently used tool to examine associations between a dependent (outcome) variable and one or more independent variables. The resulting model enables prediction of an unobserved outcome based on the observed independent variables. In rehabilitation research the dependent variable is quite often dichotomous, i. e. having just two parameter values (e. g. capable of work: yes/no). For such an outcome variable, the logistic regression model can be applied, having specific advantages in interpreting the model parameters with respect to risk factor analysis. In this paper the basics of the logistic regression model, interpretation of the model parameters and special aspects of modelling are presented. Subsequently the logistic regression model is applied to an example dataset for estimating the risk of early retirement after inpatient rehabilitation.