BACKGROUND: Personal health records are web-based applications that allow patients to directly enter their own data into secure repositories in order to generate accessible profiles of medical information. OBJECTIVE: The authors evaluated a variety of user interfaces to determine whether different types of data entry methods employed by Personal health records may have an impact on the accuracy of patient-entered medical information. METHODS: Patients with disorders requiring treatment with thyroid hormone preparations were recruited to enter data into a web-based study application. The study application presented sequences of exercises that prompted free text entry, pick list selection, or radio button selection of information related to diagnoses, prescriptions, and laboratory test results. Entered data elements were compared to information abstracted from patients' clinic notes, prescription records, and laboratory test reports. RESULTS: Accuracy rates associated with the different data entry methods tested varied in relation to the complexity of requested information. Most of the data entry methods tested allowed for accurate entry of thyroid hormone preparation names, laboratory test names, and familiar diagnoses. Data entry methods that prompted guided abstraction of data elements from primary source documents were associated with more accurate entry of qualitative and quantitative information. CONCLUSIONS: Different types of data entry methods employed by Personal health records may have an impact on the accuracy of patient-entered medical information. Approaches that rely on guided entry of data elements abstracted from primary source documents may promote more accurate entry of information.