Much of the health information available to consumers on the Internet is incomplete, out of date, and even inaccurate. Seals of approval or trustmarks have been suggested as a strategy to assist consumers to identify high quality information. Little is known, however, about how consumers interpret such seals. This study addresses this issue by examining assumptions about the quality criteria that are reflected by a seal of approval. This question is of particular importance because a wide variety of quality criteria have been suggested for online health information, including core aspects of quality such as accuracy, currency, and completeness, proxy indicators of quality such as the disclosure of commercial interests, and indicators that reflect the quality of the site or the interaction it affords, such as the availability of a search mechanism. The results of this study suggest that seals of approval are assumed to certify information quality primarily with respect to core quality indicators, aspects that subjects both consider to be important and feel relatively less able to evaluate for themselves (compared to their ability to rate proxy indicators of information or indicators of site or interaction quality). This assumption is largely inconsistent with practice: most seals of approval involve assessment of proxy indicators of information quality, rather than direct assessment of content. These results identify a problem that certification or accreditation bodies must address, since unless and until consumer expectations are congruent with evaluation practice, seals of approval seem to promise more than they deliver.