Global trade models require consolidated data with identical bilateral trade flows matrices, with exports and imports matching trade flows and closed market balances. A new methodology was developed and tested to consolidate trade matrices and market balances for more than 200 countries, around 140 products and years 1990-2004. A Bayesian Highest Posteriori Density (HPD) estimator was used in a two step procedure: first trade flow notifications of importers and exporters are rendered consistent and complete, if necessary using back- or forecasting. This step includes a so-called “revealed data quality indicator” which ranks the quality of the trade notifications per country and product. Second the aggregated totals from the estimated trade flows are used as constraints in an estimation giving complete and consistent time series of market balances. This step expresses the precision of information regarding the various elements of these balances in terms of a priori standard deviations entering the HPD estimator.