A new method for estimating total housing returns at a local level is introduced. A unique and comprehensive dataset on housing price transactions for the Paris administrative region over the 1996-2008 period is combined with sets of individual panel data collected by the NSI for a survey on rents and current expenditures used in the compilation of the French CPI (Consumer Price Index). A joint analysis of the determinants of rent level (with current and capital expenditures) and vacancy duration can be carried out with this rich set of data. Physical attributes as well as precise spatial localization are available for both prices and rents database. Hedonic regressions on rents and prices are run and an estimation procedure of the duration of vacancy spells is proposed. This information can be used to produce by imputation an estimate of the total (capital and cash flow) returns of each individual property. The huge size of the transaction price data base enables an accurate measurement of total returns for the Paris housing market. In particular, this approach does not suffer from the ìsmoothingî problem that occurs with valorisation based real estate returns indexes. Finally, the spatial and temporal profiles of returns are explored and an empirical analysis of idiosyncratic risk, with a breakdown of the specific contributions of price, rents and vacancy risks, is performed.