Purpose of the paperThis research paper has two objectives. The first is to shed light on the consistency in and quality of the applied valuation models. The second objective is to analyse uniformity on important valuation input variables throughout 1994‐2002.Design/methodology/approachMore than 150 original valuation reports are retrieved and qualitatively checked on model consistency, for example on discounting methods. The impact of the inconsistencies on the end value were calculated by using a dummy discounted cash flow model (DCF). The uniformity of the input variables net yield, discount rate and exit yield are quantitatively determined: is there a decreasing standard deviation through time?FindingsThere appears to be little consistency: the Dutch appraisers use a variety of methods within the DCF method. Cash flows are discounted quarterly in advance, yearly in arrear and averaged over the year, only three of the ten most frequent used appraisers use a flexible inflation scenario, etc. These different approaches can have a large impact on the appraisal value. As for the uniformity, the standard deviation for all three variables has not decreased through time.Practical implicationsThe conclusions and recommendations of this research have been used by the valuation committee of the ROZ/IPD Netherlands Property Index to improve and extend the valuation guidelines.Originality/valueValuation models, which are the foundation of benchmarks, have never been researched on a large scale due to confidential issues. This research appears to be the first to actually analyse valuation models of many different appraisal companies in one country, The Netherlands. The participants of the ROZ/IPD Netherlands Property Index own 85 per cent of the €38 billion institutionally invested value in real estate in The Netherlands. Their policy decisions are partially based on the comparison to the Dutch benchmark. Therefore consistency and uniformity of the valuation models is critical.