A central prediction of information economics is that market forces can lead businesses to voluntarily provide information about their quality. The mechanism that leads to this process (known as unraveling) is simple – when businesses do not disclose information, customers will infer that the information was unfavorable. Given this, why is there so little disclosure in the field? In this paper, we provide evidence of a fundamental failure in the logic of unraveling – customers may not be sufficiently skeptical about non-disclosed information. In a series of laboratory experiments, we implement a simple two-player disclosure game in which senders can verifiably report their type to receivers, and we find persistent failures of information unraveling: senders do not fully disclose their type and receivers are not fully pessimistic about nondisclosure. This suggests that even in field settings where all the standard conditions are met, one should still not expect to see unraveling. However, we are able to change the rate of voluntary disclosure in our experiment through an intervention in which we provide receivers with summary information about the disclosure decisions of past senders.