This research explores the boundaries of a valid application of contingent valuation as a method to value environmental changes. Recent criticism from social psychologists refers to unrealistic cognitive demands upon respondents. Criticism from economists point to the problem of measuring preferences in relation to non-use values. In this research the hypothesis is tested that the CV method in valuing goods with non-use value and with a complex content of information produces less valid results than with use values and non-complex content of information. Validity is tested across goods (sample survey: N = 832). Results show that scope validity is ambiguous for goods with high content of complexity. Construct validity is not guaranteed for environmental changes with non-use values: Economic factors are not significant, only attitudes are relevant. Given the importance of WTP figures in cost-benefit analysis these results are of paramount importance for the trust in contingent valuation.