As a potential alternative to current wet-lab technologies, DNA sequencing-by-hybridization (SBH) has received much attention from different research communities. In order to deal with real applications, experiment environments should not be considered as error-free. Previously, under the assumption of random independent hybridization errors, Leong et, al. [9 J presented an algorithm for sequence reconstruction which exhibits graceful degradation of output accuracy as the error rate increases. However, as the authors also admitted, a notable downside of their method is its too high computational cost. In this paper, we show that the poor efficiency of  is due to its mixing-up of situations with widely different characteristics and treating everything in the safest but also slowest way. Our new algorithm addresses this problem and pushes analysis down to a finer level where a more effective solution is proposed. As demonstrated by experimentations on real human genome datasets, this new methodology yields significant performance improvements and at the same time guarantees almost the same degree of output accuracy.