We analyze the impact of an individual's tendency to worry on willingness to pay (WTP) for a protective measure. We report on the results of a controlled experiment with real objects at stake. Worry was measured with the Worry Domains Questionnaire, an instrument determining an individual's tendency to (non-pathological) worry. Although the loss probability was relatively high and exactly specified, we find that the tendency to worry has in general a positive effect on WTP for protective measures. However when the objects at stake are given to our respondents, high worriers significantly reduce their WTP for protection whereas low worriers are unaffected. We call this tendency of high worriers a safety illusion and relate it to findings on automatic self- regulation, transitional objects in childhood, and studies on the illusion of control. High worriers appear to make more use of all these mechanisms than low worriers because they have more experience in easing their anxiety.