In 1992 the National Oceanic and Atmospheric Administration (NOAA) convened a panel of prominent social scientists to assess the reliability of natural resource damage estimates derived from contingent valuation (CV). The product of the panel's deliberations was a report that laid out a set of recommended guidelines for CV survey design, administration, and data analysis. This paper focuses on one of these guidelines -- the Panel's call for the "temporal averaging" of willingness-to-pay (WTP) responses obtained from CV surveys as one method for increasing their reliability. The panel suggested: "Time dependent measurement noise should be reduced by averaging across independently drawn samples taken at different points in time. A clear and substantial time trend in the responses would cast doubt on the 'reliability' of the finding." The purpose of this paper is to examine the temporal reliability of CV estimates. Our findings, using a CV instrument designed to measure willingness-to-pay for a program to protect Prince William Sound, Alaska from future oil spills, like the Exxon Valdez spill, exhibited no significant sensitivity to the timing of the interviews. For two samples involving independent interviews taken over two years apart, the distribution of respondents' choices "for" and "against" the protection program did not differ.