Abstract: The fitting method of selection of parameter weights is an approach to solve the ill-posed problem of geodesy,which is the reform of the Tikhonov regularization algorithm.The general formula that is used to compute the bias of the regularization resolution is deduced and the basic principles and theory of the fitting method of selection of parameter weights are introduced firstly.And then an important property is derived: the other parameters are also unbiased if the constrained partial parameters are unbiased.It indicates the result that is estimated by applying the fitting method of selection of parameter weights is conditional unbiased.This important property is different from that of the general regularization method and can be used to speed up the resolution of the double differential ambiguities in the fast positioning of GPS short baselines.Some experiments were adopted to analyze the impacts of three dimension baseline components on the resolution of the ambiguities.