This article examines the sensitivity of survey measures of willingness to pay for public goods. Visitors to a science museum in San Francisco were asked to provide estimates of their willingness to pay for saving seabirds from oil spills and for teaching English to immigrants under various experimental conditions. Willingness to pay was substantially reduced by a seemingly innocuous reminder about how many individuals would be affected by a tax or would be asked to contribute to a given cause. This finding, which cannot be explained by standard economic interpretations of willingness to pay, is consistent with previous studies showing that subtle changes in question order and wording can affect the nature of the responses.