Accurate threshold is the key factor for reducing both of missing and nuisance alarm rates, which is essential for the φ-OTDR distributed fiber-optic sensing system. In order to solve this problem effectively, the algorithm based on the selective average was proposed. During this proposed algorithm, a pre-selection scheme was achieved using the correlation difference method to determine a signal template. Then, the match of threshold template can be realized. Theoretical and experimental results show that by the proposed algorithm the threshold can be obtained to decrease the missing and nuisance alarm rates, and improve the signal-to-noise ratio(SNR) of the φ-OTDR distributed fiber optic sensing system. In comparison with current direct and moving average algorithm, the missing alarm rate is decreased by 2%, and the nuisance alarm rate is decreased by 5% and 3%. What′s more, the SNR is increased by 2.55 dB and 1.1 dB, respectively. In conclusion, the proposed algorithm was also useful to improve the real-time reliability of the system for the practical applications.