Influenced by complex structure and poor working conditions, the working status of satellite will have abnormal state that between normal and fault, which would cause not only noise, but also abnormal data existed in the telemetry data. However, when the traditional data filtering algorithm is used to denoising, the abnormal data as well as the related information could be lost. Therefore, a combined filtering method of telemetry data based on double tree complex wavelet and morphological filtering is proposed in this paper. To be specific, a semi-soft threshold filter function is designed to improve the filtering performance. Thus, a combined filtering algorithm is proposed. The experimental results show that the proposed method can effectively filter the noise in the telemetry data and retain the abnormal data.