<p>Risk classification is one of the mechanisms avoiding adverse selection in insurance. In life &amp; health markets, individual risk can be separated in three types: already visible, declared by individuals, and measured. Driven by changes in regulations and insurers’ practices, the use of certain information types is more and more limited. The consequences of such evolution on individual insurance decision are unclear, especially for high risk. We tackle this issue through a lab experiment. Participants explored the market more than theoretically expected, incurring higher search costs. They shared less inaccurate information than expected, favoring abandoning their investment project where insurance is needed over lying about their risk. Those previously refused by insurers were more likely to declare false information or abandon the insurance project.</p>