The iterative trimming approach for the robust myriad filter is proposed and developed. The operation of iterative trimming enables the myriad filter to reject the outliers in the original moving window and then the myriad is estimated on the base of the reduced set of samples. The action of the reduction of the data set is performed after estimation of the some threshold level. All samples which values exceed the threshold level are rejected. Consequently the new myriad value is calculated using the fixed point method. Compared to the classical approach of the myriad filter computation and the iterative arithmetic mean filter, the proposed modified myriad filter suppresses the impulsive noise more effectively. The numerical experiments on synthetic data and real biomedical signals allow to verify the usefulness of the proposed approach.