MapReduce is the most popular platform used in cloud computing for large-scale data processing. Generally, data processing involves multi-way Theta-joins join operations.Although multi-way Theta-joins could be processed in MapReduce by using a sequence of MRJs MapReduce Jobs, it would lead to high cost of I/O due to the storage of intermediate results between two sequential MRJs. Thus, we focus on the performance improvement of multi-way Theta-joins by reducing the number of MRJs. In this paper, a multi-way Theta-join is processed in only two MRJs, since it is decomposed into a non-Equi-join and a multi-way Equi-join and each join operation is processed in one MRJ. Our experiments show the good performance of our method.