While personal health records (PHRs) carry an array of potential benefits such as increased patient engagement, poor usability remains a significant barrier to patients' adoption of PHRs. In this mixed methods study, we evaluate the usability of one PHR feature, an intake form called the pre-visit summary, from the perspective of cognitive load using real cardiovascular patients in a natural setting. A validated measure for cognitive load, the NASA Task Load Index, was used along with retrospective interviews to identify tasks within the pre-visit summary that increased participants' cognitive load. We found that the medications, immunizations, active health concerns, and family history pages induced a higher cognitive load because participants struggled to recall personal health information and also due to user interface design issues. This research is significant in that it uses validated measures of cognitive load to study real patients interacting with their PHR in a natural environment.