There were results of performance evaluation of page segmentation algorithm on English dataset in the past. However, evaluations over Vietnamese datasets have not yet to be experienced. In this paper, the authors conducted the experiments with three start-of-the-art algorithms for page segmentation: RAST, Docstrum and Tab-Stop on the PRImA and Vietnamese data set. The results show that TabStop has the highest performance on PRImA dataset, while Docstrum outperforms, the other two algorithms on Vietnamese dataset. Besides, it is seen that RAST provided the lowest results on both datasets.