Abstract: Based on the statistic detecting methods, a reasonable algorithm was put forward to reduce the sample size of turbulent experiment. Wavelet analysis method was adopted in the algorithm to get the characteristic parameters of turbulent flow in both time domain and frequency domain. Comparing with the former algorithms, the reduced data size by the algorithm is larger, but it includes the same information with the initial sample data statistically. An example was provided to prove the reliability and rationality of the algorithm, where the sample data is from the experiment on the mechanism of riblets drag reduction and the turbulent kinetic energy is selected as the statistic characteristic parameter.