In this paper we investigate the risk factors associated with real estate investment. We explore a rich database of over 100 000 transactions mainly for residential assets in the Paris area over the 1973 ñ 1998 period. The main risk factors are identified using a Principal Component Analysis as well as a Stepwise Regression Method. The first method shows that linear or log-linear combinations of factors such as interest rates, interest rate spreads, financial market return, rents, unemployment, or even market traded real estate cannot capture physical real estate return risk in a satisfactory manner. The second method indicates that if one were to nevertheless opt for a factor model, the consistent factors should be rents, unemployment, traded real estate and a population index. Comparisons of this factor model index with both a Weighted Repeat Sales (WRS) index computed for the Paris area, the IPD index and the square-foot price index, yield interesting implications concerning real estate risk, market participant behaviour, and the so-called 1990s ëspeculative bubbleí.