The present paper focuses on some important requirements for understanding patent searchreports in view of their use for statistical analysis. It is pointed out and illustrated that thecomprehensiveness and the quality of a given search report may vary significantly as a function ofthe patent office drawing up the report. These differences imply consequences with respect to thesafe use and interpretation of the data. The authors stress that a sound analysis based on patentcitation data can only be performed in a meaningful way if the analyst has a minimum knowledgeof the underlying search reports.