OBJECTIVES: Estimate the willingness to pay value for the health improvement of the general Chinese using contingent valuation method. And comparing the difference of results between on-line and face-to-face survey. METHODS: Two widely used preference-based health-related quality of life instruments were used to elicit utility. Firstly, the respondents were requested to evaluate their own health states with the instruments of VAS and EQ-5D. Secondly, collected the respondents’ WTP for obtaining an improvement in health from their own health state to the perfect health state, using an open ended method. Finally, we collected the social-demographic characteristics of the respondents. RESULTS: We randomly sampled 97 respondents in total. Among them, 61 were from an on-line panel and 36 were from face-to-face survey in July 2015. The average utility of subject’s own health were 0.96 and 0.87, based VAS and EQ-5D respectively. With the EQ-5D instrument, more than 70% (68/97) of the respondents thought the utility of their own health were 1, namely the perfect health state. For those respondents, the health improvement were zero, so the willingness to pay for health improvement cannot be estimated. On the other hand, the social-demographic characteristics in on-line and face-to-face survey showed no difference except the age characteristic. CONCLUSIONS: It is not a good idea in China to estimate the respondents’ willingness to pay for the health improvement that from their own health states to the perfect health state. We should try to explore other methods to express the health improvement, for example, assume that the respondents in a not perfect health state. We also conclude that the respondents in on-line survey were younger than the face-to-face survey. So we suggest that on-line and face-to-face survey can be used in the same time to ensure the respondents were representative.