In applied contingent valuation research, the so-called multiple-bounded dichotomous choice (MBDC) willingness to pay (WTP) question format allows respondents to express uncertainty about their valuations. Response data from such questions are typically treated as ordinal, since the response scale is based on rank-ordered verbal expressions. Their probabilistic nature in combination with a review of empirical survey results suggest, however, that WTP response data based on verbal expressions of uncertainty reflect an underlying unmeasured variable at the ratio scale level. The implications were empirically analysed using data from a recent contingent valuation study on forest land protection for biodiversity purposes. The verbal response categories to the MBDC (WTP) question were recoded as probabilities. The effects of different probability recodings were analysed when the distribution of uncertain WTP was assumed to be known. Taking uncertainty into account by means of the underlying probability of acceptance will make contingent valuation results more meaningful and transparent for both analysts and users.