This paper investigates whether reliable valuation results may be expected when valuers, in carrying out a valuation task, use an identical data set as well as an identical valuation method. The hypothesis is that under such circumstances an identical output of property values of different valuers is to be expected. Earlier work has suggested that valuations are not always as accurate as they should be. Earlier work also suggests that valuations may incorporate elements of cognitive bias, which contribute to this inaccuracy. This paper reports an empirical investigation which was undertaken with professional valuers of a well-known reputation and comprises a mock valuation task. Several professionals drawn from well-known agencies were asked to carry out two different property valuations with two different methods according to the guidelines of the Dutch ROZ-IPD Index. The valuers received identical input in the form of a complete data set of quantitative and qualitative variables. It appears from this research that despite the fact that identical input information was provided, the valuation results of these valuers show large differences. In contrast to what was to be expected the hypothesis was rejected. The paper examines the causes for the differences in value opinion, which depend on the way in which valuers process information. Implications of these findings are discussed in the paper. It is concluded that the results of this experiment may support the presence of anchoring, search limitation and data adjustments, which lead to cognitive bias.