The overwhelming basicity of all analysed surfaces strongly dependent on the choice of liquid triplet used for contact angle measurements and the negative values sometimes obtained for the square roots of the acid-base parameters can be summarized as the main problems arising from the application of the Good-van Oss-Chaudhury (GvOC) theory to the calculation of Lewis acid-base properties of polymer surfaces from contact angle data. This paper tries to account for these problems, namely: (1) the Lewis base, or electron donor component, is much greater than the Lewis acid or electron-acceptor component because of the reference values for water chosen in the original GvOC theory. A direct comparison of the acidic component with the basic one of the same materials has no meaning. A new reference scale for water which is able to overcome this problem is suggested. For the calculation of acid-base components, a best-fit approach is proposed which does not require any starting information about the liquids or polymers and can yield estimates of the acid-base parameters for both the liquids and the polymers involved; (2) the strong dependence of the value of the acid-base components on the three liquids employed is due to ill-conditioning of the related set of equations, an intrinsic and purely mathematical feature which cannot be completely cured by any realistic improvement in experimental accuracy. To reduce or eliminate the effect, one only needs a proper set of liquids, representative of all kinds of different solvents: (3) the negative coefficients appear as a simple consequence of measurement uncertainty, combined with the possible ill-conditioning of the equation set. We cannot exclude, however, that in some cases they could have a different origin.