Multiple polymerase chain reaction (multiple PCR) is one of the most important techniques in molecular biology. The selection of a suitable set of primers is very important for multiple PCR experiments. The primer selection problem is to minimize the number of primers required to amplify a set of DNA sequences. If the minimum set can be used to amplify the entire target DNA sequences, the experimental costs and time will be reduced. But the primer selection problem was proved to be an NP-complete problem. In this thesis, we develop an efficient heuristic algorithm for selecting a set of primer candidates, each may be able to amplify more than one target sequence. Those primers are called universal primers. The universal primer finding can be viewed as the local motif finding in our method. We modify the score function of the original Gibbs sampler method to find local motifs. The new score function is added a new parameter, weight parameter. The weight parameter can guide the Gibbs sampler method to find local motifs with the local view. Then, the complementary sequences of those local motifs are input into the binary integer programming. Thus we can reduce the size of the solution space. We also test our method on some artificial domains and two gene families. All the results show that we get some improvements on the problem.