The objectives of this paper are to propose and to implement for the French case an approach for the estimation of consumers' willingness to pay for tap water quality. This method is indirect and is based on the observation of the households' demands for soft drinks and bottled water. The lack of data we face and the nature of the question we have to deal with preclude direct use of the standard methods of environmental goods valuation: the "averting expenditure approach" and the "public good as quality characteristic of a privately consumed good" approach. In fact, the tap water quantities that are drunk by the households are not observed and soft drinks are not solely purchased to circumvent tap water quality problems. Also, in the first part of this paper, a specific theoretical framework is developed. It explicitly considers the trade off: drinking almost free tap water of low quality versus purchasing more expensive soft drinks and uses the "weak complementarity" and "weak substitutability" concepts. Given that the theoretical framework rests on many unobserved variables, a structural econometric model that embodies all the features of the theoretical model cannot be specified. The second part of the paper shows how the standard empirical analysis of equivalence scales can be adapted in order to overcome this problem. The remaining of the study is devoted to the presentation of an empirical illustration of the approach for France.