Reliably and fast matching fingerprints is still a challenging problem in a fingerprint verification system. This paper proposes a minutiae matching algorithm that uses minutiae-centered circular regions to help ensure the speed of matching and the robustness to non-linear distortion. In our method, a circular region is constructed around each minutia, which can be regarded as a secondary feature. Using the constructed regions, the proposed algorithm can find matched minutiae more rapidly via regional matching. Since each minutia's region is formed from only a small area of the fingerprint, our algorithm is more tolerant to non-linear distortion when compared to global matching approaches. On the other hand, the area of the constructed region is much larger than that of local neighborhood in local matching approaches, which means that circular region, including a larger subnet of minutiae, is more reliable and distinct feature. Experiment results show our algorithm 's good performance on processing speed and accuracy.