Abstract-Blind recognition of STBC is an important issue in the non-cooperative context. However, the research on blind recognition over frequency-selective fading channels has been rarely reported. This paper introduces a novel and efficient algorithm for the frequency-selective fading channels to classify STBC via Kolmogorov-Smirnov (K-S) test when a single antenna is employed at the receiver. The proposed algorithm exploits the discrimination features provided by the empirical cumulative distributions (CDFs) of the received signal. The distance between CDFs is employed relies on the two-sample Kolmogorov-Smirnov test. The algorithm's performance is compared under the condition in different modulation mode, sampling factors and confidence interval. Monte Carlo simulations are conducted to demonstrate the well performance of the proposed algorithms, and the proposed algorithm does not need the estimation of channel, noise statistics and modulation type, and, consequently, is well-suited for non-cooperative communication context.