Through "health risk calculator" websites, Internet users can obtain personalized and interactive predictions about health risks. While it seems appropriate to provide laypersons with educational health information, prior research emphasizes the importance of understanding behavioral responses when communicating risk information. In order to systematically explore and analyze these relationships in the context of online health information, a new diabetes risk calculator website was developed to serve as an intervention in which the presence of personalization and interactive feedback could be manipulated for randomized experimentation. The website was integrated with pre- and post-intervention surveys in order to assess users in terms of information usage and risk perceptions. In two preliminary experiments, there was a small, unexpected negative impact of personalization on multiple measures of information usage. The implications of these results are discussed in the context of improving the presentation of online risk information for practical and experimental evaluation purposes.