Building Information Modelling (BIM) processes imply an intensive use of information. Nevertheless, several studies revealed critical issues in the data quality of information models. While some studies presented interesting works in the evaluation of model quality with reference to IFC. The analysis of the data quality issues in native models remains a research gap as well as the understanding of where these issues are generated. This research proposes an analysis of four information models to evaluate and classify data quality issues according to three dimensions, ie accuracy, coherence and completeness. Results highlighted user behaviours and/or technological limitations in real-world applications.