Compared with other test generation methods, the problem of automatically generating executable test cases is addressed in this paper. First, the existing test generation methods for EFSM--specified systems are also presented. These methods combine both control and data flow techniques, but the executability problem has not been solved. For this purpose, a methodology which will solve, if not completely, partially the executability problem and mainly the problem of including all the influencing loops in the text sequences and finding how many times an unbounded loop must be executed is proposed in this paper. In addition, the detailed algorithms are presented.