A research was made on two flutter boundary prediction methods: flutter margin method and auto-regressive and moving average model (ARMA) stability analysis method. The influence of parameter identification errors on the accuracy of boundary prediction was analyzed through numerical simulation case. And the prediction effectiveness of these two methods was analyzed by two wind tunnel flutter tests. The results showed that: (1) the identified damping error had less influence on the criterion of flutter margin method than the identified frequency error

and ARMA stability analysis method was more sensitive to the identified damping error than flutter margin method contrastively

thus the flutter margin method had better robustness than ARMA stability analysis method. (2) For wing bending-torsion coupling flutter

these two criterions had approximate variation curves

which differed in order of magnitude approximating to the biquadrate of the sampling frequency

and good decreasing trends helpful to predict flutter boundary in advance were presented. (3) For in-plane bending flutter

ARMA stability analysis method may have better applicability than flutter margin method.