Abstract: In this paper, we propose a new type of estimator called pre-test estimator （PTE） for seemingly unrelated linear regression system based on the least square estimator （LSE） and the covariance improvement estimator（CIE）. The statistic used to measure the correlation between sample information and additional information is obtained. We also discuss some properties of PTE and make comparison with LSE and CIE in terms of mean square error. Furthermore, Monte Carlo simulation is reported to show the performance of PTE.