According to the characteristics of the reflection spectrum of Fiber Bragg grating(FBG),the influence of power threshold value upon five kinds of typical peak-detection algorithms was studied,and the importance of threshold optimization for reducing peak-detection errors is clear and definite. The threshold value of each algorithm was optimized,and the five algorithms were compared and analyzed from the error of peak-detection and the operational efficiency by selecting different sampling intervals. Experiments show that the relationship between each peak-detection algorithm and power thresholds are different,select the appropriate power threshold for each peak-detection algorithm can reduce their error effectively,improve the resolution and accuracy of wavelength of FBG sensing system. Except centroid detection algorithm,increasing the sampling interval does not improve the operational efficiency of each algorithm significantly. Gaussian fitting algorithm has the smallest error and the best stability,and is appropriate for the demodulation of static signal and low-frequency dynamic signal. Centroid detection algorithm has the highest operational efficiency and relatively small error,and can be used in the demodulation of intermediate-frequency and high-frequency dynamic signal.