Flight Safety Margin, based on the situation of flight and from an operation point of view, provides a new tool whereby flight safety can be analyzed numerically. The flight operation is viewed moving on a virtual terrain in the abstract situation space. Any normal real flight will thus be delineated by a time-varying continuous curve around the centerline defined by the standard flight condition. The flight safety margin describing how far the present flight situation is from the accident boundary is de-fined as the inverse of the needed performance of the crew to recover the present situation back to the standard condition and scaled from zero to one. A questionnaire is designed to measure the perceived needed performance. With the chosen situation parameters as the inputs, the surveyed results are then converted to the flight safety margin, representing the outputs of the training examples. The expert sys-tem using neural network can thus provide the quantitative flight safety margin given situation parameters from real flight condition. The present methodology has been tested with the FOQA data from final approach in real cases including the Nagoya and Da-Yuang accidents. Meaningful results are obtained although there is still much room for improvement.