There are several methodologies for collecting data for semantic distances between word concepts. Most methods obtain information reflecting only declarative knowledge. Few if any of these methods attempt to capture both declarative and procedural know ledge present in the internal mental representations. The present study attempted to examine the structure of declarative and procedural knowledge in phone and automatic teller machine (ATM) commands. Three different types of statistical analyses were c ompared: multidimensional scaling (MDS), hierarchical cluster analysis (HCA) and pathfinder network analysis. The results show that hierarchical cluster analysis generated the most comprehensible and coherent solution for the phone commands, while none of the statistical analyses generated an acceptable solution for the ATM commands. It may be possible that procedural knowledge is not only difficult to express verbally, but also difficult to state explicitly or describe formally. Some previous studie s indicate that users may not be able to easily give very accurate information regarding their procedural knolwledge of a system.