The Maslach Burnout Inventory (MBI; Maslach, Jackson, & Leiter, 1996) is the most commonly applied measurement instrument for the assessment of burnout. In order to gain further insight into the controversy surrounding its psychometric properties, we conducted a comprehensive, robust, item response theory based psychometric analysis of a version of the German translation of the MBI within a sample of physicians recruited via a sampling procedure from a full medical register. The psychometric analysis is based on a two-step procedure which incorporates a general nonparametric analysis followed by a parametric analysis. The analysis shows departures of the MBI scales from the usually reported three-dimensional factor structure and highlights two potentially misfitting items. Deviations of the subscales from a graded response model are assessed via the use of posterior predictive p-values and results of a simple nonparametric method for scale construction are reported.