Voltage instability predictor（VIP） based on PMU is a method for analyzing voltage stability using local measurements and it employs the least squares fitting method to track the Thevenin parameters. The application of VIP is limited by the application conditions of curve fitting technique and the real meaning of the static voltage stability. The equivalent model is invalid if the input data set cannot meet appropriate limitation， and as a result the voltage stability situation will be misestimated. In this paper the underlying cause of the parameter drifting is analyzed and validity criterion is set up to determine whether the result of VIP is significant. Tests based on real electrical network operation show that the criterion is reliable.