Purpose - This paper discusses the most recent and future developments in corporate real estate management (CREM) for multinational corporates - focussing on organization structures and role definitions. The main purpose of the analysis is to show best practice business models based on business cases. Varying degrees of vertical integration of real estate management functions (portfolio, asset, property and facility management)are presented and analysed with regard to the generated value added and leverage opportunities.Approach - The presented CREM business models are based on ICME Management Consultants' project experience and on acknowledged business standards. The most important trends in international CREM were deduced from a recent survey among CREM executives conducted by ICME. Findings - The challenge of an increasing internationalization of core businesses will be the main driver for the future trends in CREM and corporate FM. CREM will mainly be organized as a global center of excellence providing a global strategic network of CREM entities. Pressure on companies and on efficiency of real estate management increases further in the course of globalization of core business activities and increasing user reuirements. For this reason, the separation of owner functions (CREM) and operation functions (FM) cannot longer be maintained. It is expected that CREM and corporate FM in globally operating companies will increasingly merge together in future. The main challenges in managing corporate real estate are increasing quality demands and space requirements of users, indicating a need for focusing on corporate FM services. The most important leverage in CREM activities will be the divestment of company sites in order to enhance core business yields by releasing real estate capital. Value of research - It is anticipated that the outcomes of the paper will assist senior executives and CREM responsibles in the definition of appropriate CREM strategies. Further, optimization potentials with regard to CREM business models can be identified.