The criticism of Contingent Valuation Methods (CVM) resolved mainly from two aspects: the validity and the reliability of the results, and the effects of various biases and errors. The recent studies addressed the validity and reliability of CVM result were arising out of different kinds of biases and errors, which included the disparities between WTA(willingness to accept) and WTP (willingness to pay), strategic effect, scope effect, elicitation effect, hypothesis effect, sequencing effect and information effect etc. A set of comprehensive methods to reduce biases and errors was used to evaluate the economic benefits of ecosystem restoration of Yangtze River Basin in Nanjing area in this paper. The nine operating approaches were (i) to apply the WTP format rather than WTA format to elicit the individuals preference in the survey to avoid the biases of income effect, and to substitute effect and endowment effect as well; (ii) to take dichotomous choice questionnaire (DCQ) as a questionnaire-guided methodology to minimize the strategic bias and obtain a powerful goodness-of-fit tests, in addition, to use both open questionnaire and payment-card questionnaire as pre-survey methodologies to provide lower and upper valuation boundary estimates for DCQ survey, and the lower bid was selected almost 100% acceptance while the upper level was almost 100% rejection; (iii) to define the evaluation object of "ecosystem restoration of Yangtze-river Basin " as a familiar, specific and clearly understanding of "water quality improvement", and to pretest the photographs of swimming in describing the degreed of water quality improvement, so the part-whole problem ( scope bias) would be minimized where respondents have the experience of valuing the good in question;(iv) to focus on individuals' intention rather than attitudes toward their future payment behavior, to analyze the respondents' intention to determine whether their answers really signified zero WTP or made in jest or protest, and to elicit the reasons for yes/no answers for the valuation question; (v) a survey of stratified random sample of 300 used in open questionnaire, the same size in payment-card questionnaire, and a larger sample of 1700 in DCQ questionnaire. All the sample sizes were significant; (vi) The trained investigators undertook the individuals by personal interviews rather than mail survey and telephone interviewing; (vii) to chose tax and higher-water-price as the realistic, plausible, appropriate and forced payment vehicle, and to remind the respondents that their choice would be the base for Nanjing's 12th five-year urban water price reform plan, which could reduce the moral bias significantly; (viii) The distribution function of WTP was made by median-estimation model in stead of mean-estimation model, because the former model had better robustness and underlying the economic implication that the WTP for a good was constringed by income; (ix) to combine double-bounded DCQ rather than single-bounded DCQ, log-log model rather than logic model to estimate the WTP value, because double-bounded DCQ was statistically more efficient than the single-bounded DCQ, and log-log model resulted in higher validity and reliability. The individual WTP elicited in this study was 270.7￥ RMB per household per year. The result provides a scientific base for making the Nanjing's 12th five-year urban water price reform plan.