In order to objectively quantify the flight safety resilience of airlines, a QAR data based flight safety resilience assessment method (QAR-FSRAM) is proposed. Firstly, the concept of flight safety resilience is defined through theoretical analysis, and its characteristic manifestations in flight performance data are described. Subsequently, based on QAR data, the QAR-FSRAM is constructed using the Short Time-series Expression Miner (STEM) technique. Finally, an empirical evaluation is conducted using the flight data of Airline A during the approach phase as a case study. The results show that the performance data characteristics reflecting flight safety resilience are characterized by two dimensions: trend and deviation. Four standardized steps are involved in the QAR-FSRAM: construction of flight resilience performance characterization variables, quantification of their trend characteristics, calculation of their deviation characteristics, and analysis of the airline’s overall flight performance structure. It is revealed by the findings that relatively low flight safety resilience is exhibited by Airline A during the approach phase, with a calculated resilience index of only 0.51. The effectiveness of the proposed QAR-FSRAM method is validated through a comparison between the resilience assessment results and Flight Operations Quality Assurance (FOQA) exceedance detection.