New objective and accurate feedback detection method, transfer function variation criterion (TVC), has been developed for evaluating the performance of feedback cancellation techniques. The proposed method is able to classify stable, unstable, and sub-oscillatory stages of feedback in hearing aids. The sub-oscillatory stage is defined as a state where the hearing aid user may perceive distortion of sound quality without the occurrence of oscillation. This detection algorithm focuses on the transfer function variation of hearing aids and the relationship between system stability and feedback oscillation. The transfer functions are obtained using the FIR Wiener filtering algorithm off-line. An anechoic test box is used for the exact and reliable evaluation of different hearing aids. The results are listed and compared with the conventional power concentration ratio (PCR), which has been generally adopted as a feedback detection method for the performance evaluation of hearing aids. The possibility of real-time implementation is discussed in terms of a more convenient and exact performance evaluation of feedback cancellation techniques.