Abstract: The problem of gross error detection and identification becomes the bottleneck of data reconciliation A new approach to gross error identification based on the reliability, precision ranking of the flowmeters as well as the residual of single node is presented It is assumed that the measurement error independently tends toward normal distribution and the reliability of every flowmeter is no lower than its lower limit The distribution of probability density for different situations of gross error at a single node is edfined The computation procedure for the different probabilities is given by using reliability and precision information of the instrument in/out a node and the residual of the node. Then the magnitude of different probabilities can be used as the criteria for gross error identification Two simulations are given and a comparison is made between the new approach and previous researches It is shown that the proposed method is quite effective in gross error identification especially when the system comprises of a relatively large number of flows