The use of traditional trade-weighted average tariffs biases the estimated welfare benefits of trade reform downwards through averaging distortions and use trade weight bias. Decomposition of total tariff variation suggests that over half the variation in EU agricultural tariffs is lost in standard models, a loss compounded by weighting bias. The Bach-Martin aggregation procedure is used to avoid this loss of information in an analysis of EU and Indonesian tariff reform. The results increase the estimated global benefits of EU agricultural trade reform by over 150%. Inappropriate aggregation may be causing very substantial underestimation of the global gains from agricultural trade reform. For Indonesia, the underestimation of the benefits using a weighted average approach is much more serious for nonagricultural products, where the variation in tariffs within groups is more important.