Niagara Mohawk Power Corporation utilized a demand revealing public good mechanism to implement a green electricity program for provision of renewable energy and planting trees. This GreenChoiceTM program provided an opportunity to test the reliability of contingent valuation for predicting actual participation levels. In this study, participation levels predicted by hypothetical open-ended and dichotomous choice questions are compared to a reference level obtained from the actual GreenChoiceTM program. This approach represents an important improvement over past public goods contingent valuation validity tests which have relied on voluntary contribution mechanisms to elicit actual willingness to pay, and thus are likely to overestimate hypothetical bias because of free riding. Yet, even with a demand revealing mechanism and controlling for awareness, hypothetical participation levels obtained from dichotomous choice responses are found to significantly exceed actual contributions. In contrast, open-ended responses predict actual contribution levels, in that hypothetical open-ended responses are not significantly different from actual responses. Calibration of hypothetical responses is also explored.