The Model-Driven Architecture has been launched in 2001 by the OMG. Since then, model-driven engineering has been embraced by the research community but less than hoped for by practitioners. To ensure the relevance of a research agenda, we need a good understanding of practitioners' problems, in particular with modelling. We therefore performed a literature review on the state of practice in the use of modelling languages for software engineering in the last 5 years according to Kitchenham's guidelines. This paper serves as orientation within the research field and as a basis for further research. It contributes to literature by focusing on papers discussing practical use of modelling languages and the benefits and problems perceived by practitioners. The main finding presented in this paper is that while practitioners experience benefits of modelling for analysis and design, requirements engineering, quality management, implementation and deployment, they still struggle with external tool integration/model transformation & export, cognitive fit, visual expressiveness, high effort required in acquiring skills, automated analysis and high effort required in using tools. Other findings are that modelling is mostly used for documentation and requirements elicitation, the most used modelling language is UML.