This paper introduces a type of open-ended valuation question where respondents state their willingness to pay in the form of an interval rather than a point estimate. Allowing the response to be expressed as an interval has advantages compared to traditional valuation questions: it captures potential valuation uncertainty, facilitates interpretation of uncertainty and most importantly, provides a richer set of information about individuals' preferences. Furthermore, an open-ended willingness to pay format has advantages if a survey is carried out in more than one country. Too little is known about valuation uncertainty to represent willingness to pay only as an exact value. Therefore, this value should be complemented by upper and lower boundary estimates. In this paper I present new methods for estimating these different values. The methods are illustrated with results from a survey concerning wild salmon in the Vindel River, northern Sweden. The results suggest that the upper and lower boundaries provide a kind of confidence interval for the willingness to pay, which is encouraging for estimating these values to characterise willingness to pay. The results also illustrate that some earlier criticism in the literature of open-ended questions does not apply to the question presented in this paper.