Many studies have found that clinicians require access to medical knowledge sources while viewing data in their patients' health records. To address this need, we have been developing links to various on-line health information resources from our Web-based clinical information system (WebCIS). Clinicians can access these resources through a 'Health Resources' Web page or use context-sensitive 'infobuttons' that automatically link to specific resources. To assist in understanding how clinicians interact with these two mechanisms, we have been exploring the use of an automated approach to answer questions about usage for guiding the further development of links to resources. In this paper, we discuss the use of a knowledge discovery technique (CIS Usage Mining) to obtain usage statistics from log files that record interactions with health resources and infobuttons in WebCIS, describe a Web-based interface developed to present the results, and summarize how the results could be applied. The availability of effective links to on-line medical knowledge sources from the patient health record can assist in resolving information needs, potentially improving patient care and decision-making.