Contingent valuation surveys have tended to yield results that seem to go contrary to what is standardly seen as 'rational choice'. We argue that some of the inconsistencies arise because bids for public environmental goods in contingent valuation surveys are often (at least partly) motivated by moral considerations and ethical beliefs. We analyse the expected results of CV surveys given the existence of such ethical motivations, including the valuation of actions as well as states. It is found that we cannot expect bids made on this basis to reveal preferences which obey the rules commonly assumed in the theory of consumer choice. The two standard reactions to these anomalies have been to attack the validity of the method, or to urge greater rigour in survey design and application. By contrast, we conclude that the usefulness of the method for evaluating options concerning environmental public goods depends critically on the problem definition. The method should be used for the direct evaluation of realistic policy packages, rather than to try to extract abstract values for invaluable goods. The conclusion accords with current trends, including the move towards the use of deliberative methodologies.