To be widely accepted by physicians, computerized support systems for medical decision-making must be seamlessly integrated into the clinical workflow. These systems must present accurate information in a useable format to the right person at the right time [1]. To build the large, clean, and reusable clinical databases needed for predictive modeling and for the coming wave of translational laboratory research, researchers and clinicians need more efficient medical database designs. We developed a database organized chronologically as date-variable-value entries. By collecting complete disease and treatment history for patients, the database could serve both clinical and research needs. A prostate cancer database was prototyped in Microsoft Access and subsequently upsized to a SQL Server and ColdFusion web application; the system, an open source application, is now in use at Memorial Sloan-Kettering Cancer Center and at several outside institutions. The system has reduced the difficulty of data entry, eased error checking, and improved the flexibility, efficiency, and reproducibility of creating research datasets from raw data. For clinicians, the system improves efficiency chiefly by displaying a single screen intuitive summary of patient data and by reducing the need for dictation. Currently, we are working to expand the system to other forms of cancer, and to implement a physician Tablet PC interface.