As we all know, QAR (Quick access recorder) data take notes on aircraft settings and are indispensable tools for diagnosing faults of aircraft systems. In this paper, the civil aircraft engine start monitoring domestic and abroad are studied, find out the necessary sensors in engine start monitoring system. Through a work called FMEMA (Failure Modes, Effects and Monitoring Analysis), this paper defines specific indicators for each systems involved in engine start sequence. And then, this paper builds normalized indicators according to extern conditions. The next step is building some meaningful signatures, which are computed from least square minimization process. On the basis of the above theoretical study, using Mahalanobis distance, do health assessment for each engine start sequence. This paper focuses on how to use QAR data to propose some precautions to maintenance department and engineers. Practice shows that making good use of QAR data can improve the accuracy of the health assessment for civil aircraft engine start system and the efficiency of engines troubleshooting.