Although hospital information systems (HIS) have recently become more widely used, data stored in HIS databases are not fully utilized in clinical research. One major reason is the lack of suitable tools to utilize the stored data for clinical research. Patient eligibility criteria for clinical studies often consist of multiple conditions incorporating logical operations across multiple data tables. These are too complex to perform using the existing HIS interfaces, which were originally designed for clinical business rather than clinical research. Moreover, these criteria sometimes require importing patient lists from external data sources not stored in the HIS databases. Certainly end-user computing (EUC) is effective in satisfying these demands, but in reality, EUC is too difficult for many clinical researchers to master because it requires them to understand the complicated composition of the data tables in the HIS database. Thus, we have developed a search system to meet these needs using user-friendly interfaces. Since the objective of a clinical study is to prove a relationship between a certain purported cause and the outcome, target patients must be selected based on several items of clinical information. Patient eligibility criteria for clinical studies are usually complex, consisting of multiple conditions combined with logical “AND”, “OR” and “NOT” operations as well as parentheses. It is a challenge to design user interfaces that can specify logical operations and parentheses among multiple search conditions. Our search system is designed to support set operations between the lists of search results as well as functions for loading and saving multiple lists, instead of logical operations and parentheses among search conditions. The system was developed using a middleware package for EUC, and was released in our institute. In this study, both the proposed search system and search interfaces for our institutional HIS were evaluated using application forms submitted to the Informatics Department requesting search and export of data stored in the HIS database. The requests were processed using both systems, and the results were recorded and analyzed for evaluation. Our search system fulfilled a significantly larger number of requests than the HIS interfaces (paired t-test, p<0.05).