Spreadsheets are widely used as decision support systems. However, as numerous studies have shown, many of them contain errors. A major reason for this is the lack of an explicit design of the data mod-el, which at the same time ensures flexibility and is one of the defining features of the spreadsheet paradigm. Therefore, in recent years many approaches to model-driven spreadsheet development emerged addressing this issue. In this paper, we propose a meta model which defines how the semantic models of today’s spread-sheets can be described. Thereby, it builds on the foundations of related approaches to model-driven spreadsheet development, and also captures semantic patterns of spreadsheets typically occurring in practice. Based on this semantic meta model, novel spreadsheet solutions can address concrete se-mantic patterns and thus reduce the risk of errors on the one hand, and improve the usability of spreadsheets on the other hand.