International audienceThis paper looks at a fundamental issue in Information Systems (IS) research methodology, that of the use and construction of theories. Theory usage and construction are required for developing science, however they are poorly understood and pose difficult challenges to the researcher. To explore this issue, we propose a set of criteria and apply it to a sample of papers in the field of ERP research. Our analysis highlights the preponderance of descriptive and explanatory theoretical proposals, which do not attempt to make predictions. Furthermore, as most of these studies are impossible to replicate, theory confirmation is a challenge. This work is a contribution to better understand knowledge building in IS research