Hedonic price modelling attempts to uncover information on the determinants of prices - in this case the prices are those of houses in the Greater London area for the period between 1980 and 1998. The determinants of house prices can include house attributes (such as size, type of building, age, etc.), neighbourhood attributes (such as proportion of unemployed people in the neighbourhood or local tax rates) and geographic attributes (such as distance from the city centre or proximity to various amenities) (Orford 1999). Almost all applications of hedonic price models applied to housing are in the form of multiple linear regression models where price is regressed on various attributes. The parameter estimates from the calibration of this type of regression model are assumed to yield information on the relative importance of various attributes in influencing price. One major problem with this approach is that it assumes that the determinants of prices are the same in all parts of the study area. This seems particularly illogical in this type of application where there could easily be local variations in preferences and also in supply and demand relationships. Hence, it seems reasonable to calibrate local hedonic price models rather than global ones – that is, to calibrate a model form which is flexible enough to allow the determinants of house prices to vary spatially. Geographically Weighted Regression (GWR) (Fotheringham et al. 2002) is a statistical technique that allows local calibrations and which yields local estimates of the determinants of house prices. GWR was recently used to investigate spatial variations in house price determinants across London separately for each of the years between 1980 and 1998 (Crespo et al. 2007). The result of the GWR analysis is a set of continuous localised parameter estimate surfaces which describe the geography of the parameter space. These surfaces are typically visualised with a set of univariate choropleth maps for each surface which are used to examine the plausibility of the stationarity assumption of the traditional regression and different possible causes of non-stationarity for each separate parameter (Fotheringham and al. 2002). The downside of these separate univariate visualisations is that multivariate spatial and non-spatial relationships and patterns in the parameter space can not be seen. In an attempt to counter this inadequacy, in a previous study we suggested to treat the result space of one single GWR analysis as a multivariate dataset and visually explore it (Demšar et al. 2007). The goal was to identify spatial and multivariate patterns that the separate univariate mapping could not recognise. In this paper we extend this approach with the temporal dimension: we use Geovisual Analytical exploration to investigate the spatio-temporal dynamics in a time series of GWR hedonic price models. The idea is to merge the time series of GWR result spaces (one space per year) into one single highly-dimensional spatio-temporal dataset, which we then visually explore in an attempt to uncover information about the temporal and spatio-temporal behaviour of parameter estimates of GWR and consequently of underlying geographical processes.