"This original study examines the potential of a spatiotemporal autoregressive (STAR) approach both in modelling transaction prices on the Paris office property market and in building a transaction-based office price index. To our knowledge, this study is the first one to apply such a method on French or European office market. The first part deals with the presentation of our data set on the Parisian office property market. The data set is from the Paris Region notary office (ìChambre des Notaires díŒle-de-Franceî) and consists of 2,551 office transaction units between the first quarter of 1990 and the third quarter of 2005. We use the exact X -- Y coordinates to sort each transaction by geographical area. In particular, we focus on the spatial specifities of the Paris Region market, with two business districts: the Central Business District (including the famous ""Golden Triangle"" where a large number of the most expensive transactions are concentrated) and the La DÈfense Business District. In the second part, we begin by using a standard hedonic method to identify the main variables explaining the movement of transaction-based office prices. We take care of the above mentioned spatial specifities of the Paris Region market. Additionally, we take into account the potential selection bias that may occur when the sample of transaction properties used for the estimation does not accurately represent the population of properties. To deal with this problem, we use probability weights estimated by the Paris Region notary office in our estimation procedure (Weighted Least Squares) to correct for unequal probabilities of selection. We then use the spatiotemporal autoregressive (STAR) approach first employed by Pace, Barry, Clapp and Rodriguez (1998), that incorporates a spatiotemporal filtering process into the conventional hedonic function. When comparing the obtained estimates from the STAR method to those of the traditional hedonic price method, we find a strong evidence of the presence of spatial and temporal interdependences among office transaction prices. Hence, the use of STAR method leads to a significant improvement of the modelling of the Paris region office market. Our results also confirm that the main additional determinants of the log of transaction prices on the Paris Region office market seem to be: the period of construction and the nature of the transaction (entire vs part of building, new vs second hand). In the third part, we build the office transaction price indexes derived from the two estimation methods (standard hedonic and STAR). We find significant differences between the two price indexes. In particular, it appears that the spatiotemporal autocorrelations lead to a substantial reconsideration of the boom in transaction prices since 1997, that seems much more pronounced than in a standard hedonic estimation."