Studies of the patterns of risk and returns in commercial property and of the application of portfolio theory in real estate typically have based their analysis on data aggregated by geography and sector. Thus, for example, UK studies seeking to identify ways of structuring a real estate portfolio have clustered IPD segment or town level data in an attempt to identify optimal groupings; other research has sought to imply individual asset volatility from aggregated data, often making heroic assumptions. However, given the heterogeneity of property assets, the behaviour of individual asset returns within those groupings may be very different. Those few studies that have used individual data have often suffered from small sample size problems. In this paper, both portfolio groupings and drivers of individual returns are re-evaluated through the application of multivariate, exploratory statistical methods to individual level property data from the IPD UK databank. The aims are to discover whether widely accepted views about the structure and drivers of the real estate market are empirically supported by evidence from individual investments; and to identify any additional dimensions in the return generating process.