ABSTRACT Multiple sequence alignment (MSA) is an essential procedure conducted in any study that compares biological sequences (DNA, RNA and protein). A variety of tasks in molecular biology, computational biology and bioinformatics depend on sequence alignment, and its quality affects the downstream analysis. The challenge in computing sequence alignments is the tradeoff between quality (i.e. accuracy) and efficiency. A variety of MSA methods have been developed and their main drawback is that the inore time-efficient the method, the lower is the accuracy of the produced alignment. Methods that produce accurate sequence alignments, known as exact algorithms, tend to have very high computational complexities and are impractical for aligning large number of sequences. Due to the significance of MSA accuracy in many biological studies and the insufficient accuracy of available MSA methods, there is a genuine need to develop a time-efficient solution that produces accurate MSA. One way to improve the efficiency of accurate methods is to deploy a number of processors working concurrently (in parallel) on the problem to reduce the computation time. In this research, an accurate parallel MSA solution is developed based on an exact pair-wise method: Needleman Wunsch (NW) algorithm. First, the NW algorithm is extended from pair-wise to multiple sequence alignment and then parallel solutions are designed using two well known parallel computing approaches: message passing and shared memory to improve the performance of our extended NW algorithm for MSA. Moreover, we conduct an analytical and experimental performance evaluations of the proposed solutions. The performance evaluation results have shown that the speedup improves substantially as the number of sequences increases. The improvement is higher for higher number of processes up to a certain limit. However, the speedup does not improve with longer sequences because the workload distribution among the processors is done across the number of sequences. Therefore, the designed parallel solutions in this study are efficiently adequate for applications that require aligning a large number of sequences accurately, such as Genome-wide association studies GWASs, using a large number of processors (such as large multicore systems or supercomputers(