Previous studies of the correspondence between prices and valuations in the UK are considered to be deficient in their formulation of the regression relationship between prices and valuations, in their statistical analyses and in their interpretation of pricing accuracy. Applying a rigorous econometric approach, this paper reports the results of more extensive testing than has been undertaken to date, thereby enabling robust conclusions. In particular, evidence is found for valuation bias under three different market environments. This supports the claim that valuers may not be using all available information at the time of valuation. Figures for pricing/valuation accuracy are reported, showing a wide range of uncertainty as regards the eventual selling price. The results imply that performance measurement and league table rankings may need to be qualified and that the routine application of fixed parameter unsmoothing procedures to valuation-based property indices may be invalid.