The wavelet de-noising approach is one of the most efficient techniques for noise reduction. Except for the mathematical elegancy in its theoretical aspect, a successful application requires appropriate determination on the decomposition level. This paper proposes a novel approach to selecting an appropriate decomposition level by applying the Gamma test on the noisy data to estimate the noise variance. A number of experiments were carried out on a set of smooth signals corrupted by white Gaussian noise to verify the effectiveness of the proposed approach and a set of traffic flow data was used to demonstrate the application of the proposed approach on the real-world de-nosing task. The experimental results indicate that the proposed approach is effective to determine the decomposition level.