This paper proposes a hybrid transform coding technique, where the intraframes of video sequence are coded by discrete wavelet transform and the interframes are coded with discrete cosine transform technique. It also proposes the selection of sequence of frames predicted from the reference intraframes. This proposal consistently minimizes the prediction error for the predicted frame for further processing. The experimental results show that the proposed hybrid transformation coding technique outperforms over conventional transformation coding technique in terms of encoding time and prediction errors.