The objective of this paper is twofold: first we use the methodology developed by Goolsbee and Klenow (2006) (G&K) that estimates the consumer's surplus of Internet's connection for France in 2005. Second, our paper challenges the initial methodology developed by G & K by using two complementary hypotheses and a Heckman's two stage estimation method. The first hypothesis take into account the concavity of the demand function of Internet and the second adds more realistic non-monetary variables. We also make some differentiation between Internet adoption and access frequency patterns using Heckman's (1976) correction procedure to resolves the selection problem. We find that French time opportunity cost is three times more important than connection cost. We find also that high-income people were more able to adopt Internet, but they spend less time online than low-income ones. In 2005, the French consumer surplus ranged between 1240$ and 3126$ if we use the G& K methodology, between 1679$ and 3126$ if we use our concave demand function, but between 2107$ and 2651$ if we use our two stage estimation method.
