The increasing use of online review sites is creating new challenges for user privacy. Although reviews are public, many users inadvertently disclose private information about relationship, location, and temporal attributes to the world. This research protects users of online review sites from the inadvertent disclosure of private information in three ways. First, the types of unstructured and structured information made public by online review sites are characterized and used to grade those sites on their attention to privacy. Second, a privacy-check tool that uses keyword matching and named-entity recognition to annotate potentially sensitive review text is presented. Third, we raise awareness of the privacy threat in online review sites through examples and statistics derived from the privacy-check tool.