A new approach is employed in GPS rapid positioning using several-epoch single frequency phase data. Firstly, the structure characteristic of the normal matrix in GPS rapid positioning is analyzed. Then, in the light of the characteristic, based on TIKHONOV regularization theorem, a new regularizer is designed to mitigate the ill-condition of the normal matrix. The accurate float ambiguity solutions and their MSEM (Mean Squared Error Matrix) are obtained using several-epoch single frequency phase data. Combined with LAMBDA method, the new approach was used to fix the integer ambiguities correctly and quickly using MSEM instead of the cofactor matrix of the ambiguities. Finally, a baseline over 3 km is taken as an example. The fixed integer ambiguities by the new approach using five epoch single frequency phase data are the same as those fixed by Bernese software using long time data. The success rate of fixing the integer ambiguities is 100 percent using 197 group data. Compared with the traditional methods, the new approach provides better accuracy and efficiency in GPS rapid positioning. So, the new approach has an extensive application outlook in deformation monitoring, pseudokinematic relative positioning, and attitude determination, etc.