Aiming at the problem of large steam flow measurement error in steam pipe network that easily causes energy waste during steam management and scheduling, this paper presents an on-line rectification method of steam flow measurement error based on principal component analysis(PCA). Firstly, the flow measurement values of the nodes in steam pipe network are filtered using PCA method to eliminate the random errors. Secondly, the squared prediction error(SPE)based on PCA model is used to estimate the loading conditions of the steam pipe network. Then, the data reconciliation method based on Lagrange multiplier is used to process the measurement errors considering the magnitude and variance of the steam flow. Lastly, the residual error matrices are used to carry out the on line rectification of the data in next statistical cycle. The presented method was applied to the steam pipe network of steel corporation, Experiment results show that after the rectification of the steam flow measurement error using the error rectification software based on the proposed method, the accumulative errors of flow measurement are decreased by 99.09%, partial measurement error is eliminated effectively, and the balance between supply and demand of steam flows in steam pipe network is achieved.