The contingent valuation method (CVM) is a direct interview approach that can be used to provide acceptable measures of the economic value of preservation of natural resources. While the estimates from CVM may not be perfect, neither are estimates from any other economic or physical science model. The need for an assessment of the nonmarket benefits of restoring Ejina banner's ecosystem is clear. In this paper, we take it as a case to analyze the WTP of restoring ecosystem service. Obtaining accurate benefit estimates using CVM requires detailed descriptions of the resource being valued. In our survey, we use a 12-page paper with maps depicting the reason why Ejina Banner ecosystem deteriorated, the means by which ecosystem services could be restoring from their current level. The actual valuation portion of the survey has three elements:⑴ portrayal of the resources to be valued,(2)description of the particular mechanism to be used to pay for the resource, and (3) the question format used to elicit the respondent's money amount of willingness to pay. Due to concern over the influence of institutional and culture setting, the method of payment we adopted is payment cards in the survey. We employed an open-ended (OE) question format and a non-parametric model to estimate the welfare of restoring Ejina ecosystem. Results from 500 in-person interviews indicate that per households would pay an average of Y44.90 per year. Y49.42 for the main river household, Y38.23 for the rest of Heihe basin. The aggregate benefit to residents of the Heihe basin is Y 1.95 million annually for 20 years. Taking into account the market discount rate, the aggregate benefit(present value) of restoring Ejina banner ecosystem is Y 12.24 million. These estimates suggest that the general public in Heihe valley would be willing to pay to restoring the Ejina Banner ecosystem. Finally, we put forward some suggestions on how to apply and improve the contingent valuation method in our next research, and believe future efforts may be able to apply CVM efficiently accompanying with benefit-cost analysis.