This disclosure relates to screening potentially inappropriate online content. A flag is received that indicates online content is potentially inappropriate. The flag is generated by a source, such as a user, or a content classifier. The potential accuracy of the flag is determined or inferred based on a variety of factors, including a reputation of the source, and the flag is categorized as requiring higher priority administrative action or requiring lower priority administrative action based in part on the potential accuracy of the flag. In addition, administrative action is taken based in part on the categorization of the flag.