A second order adaptive notch filter using an allpass filter and adaptive line enhancer has been proposed. An adaptive algorithm of this notch filter uses a gradient of mean square of an estimation error with respect to all tap coefficients of the allpass filter. There is no guarantee to the convergence of the adaptive algorithm when the estimation error is large. In the adaptive notch filter, a conventional adaptive algorithm whose the convergence is guaranteed is used till the convergence is guaranteed. Then, by using the switching condition which determines the convergence of the adaptive algorithms, the adaptive algorithms are switched. However, the convergence rate of the adaptive notch filter is decreased due to making an incorrect assessment of the switching adaptive algorithms. In order to solve the problem, an adaptive algorithm using a new switching condition is introduced to the adaptive notch filter. Finally, computer simulation results are presented to confirm the convergence characteristics.