The increasing demand for quality information systems (IS), has become quality the most pressing challenge facing IS development organisations. In the IS development field it is generally accepted that the quality of an IS is highly dependent on decisions made early in its development. Given the relevant role that data itself plays in an IS, conceptual data models are a key artifact of the IS design. Therefore, in order to build "better quality" IS it is necessary to assess and to improve the quality of conceptual data models based on quantitative criteria. It is in this context where software easurement can help IS designers to make better decision during design activities. We focus this work on the empirical validation of the metrics proposed by Genero et al. for measuring the structural complexity of entity relationship diagrams (ERDs). Through a controlled experiment we will demonstrate that these metrics seem to be heavily correlated with three of the sub-factors that characterise the maintainability of an ERD, such as understandability, analysability and modifiability.