The efficient operation of any modern metallurgical facility depends on the accurate measurement and estimation of flow rates, inventories, and composition of their intermediate and final products. All of this information is subject to both statistical errors and gross errors, which can lead to poor estimation of efficiency, yields and specific energy consumption. The detection and the elimination of these errors require not only plant data, but also product transactions and the operational events. Once the gross errors have been eliminated, it is possible to estimate validated performance indicators, such as grade recovery or yields. This paper describes the methodology to implement a unification and gross error detection system using the available infrastructure of both information and data reconciliation system. It covers the different infrastructure layers, from the data collection layer to the data validation and reconciliation layer. In addition, the impact of these methodologies on the decision-making process is presented. (C) 2005 Elsevier Ltd. All rights reserved.