This study re-examines the potential role that direct real estate can play in institutional mixed-asset portfolios. The paper examines the statistical improvement in performance that can result from the inclusion of real estate in an international mixed asset portfolio, using both in-sample and out-of-sample data. Using US real estate data the results provide evidence that in most cases real estate does not lead to a significant improvement in portfolio performance in sample. However, out-of-sample tests indicate that the asset does provide a valuable diversification asset, with significant improvements in performance relative to a base capital market only portfolio.