Various indicators (e.g., VAT-to-GNP ratio, VAT evasion rate) suggest that problems in VAT administration have not only persisted during the 1990s but also appear to have worsened. This study makes use of the 1994 Input Output Table to estimate industry benchmarks for the ratio of VAT-able inputs to total output as well as the effective VAT rate. In so doing, it also attempts to decompose the sources of the leakage in VAT collections in 1999-2001. While in the aggregate the VAT-able purchases of the firms under the Large Taxpayer Unit of the BIR are less than their total output in the current year, it appears that many of these firms do claim excessive input VAT credits. The ratio of the value of total purchases to total sales is higher the corresponding I-O based benchmark in some 68% of large VAT-payers. In this regard, the use of industry benchmarking to automatically trigger audit would help reduce the leakage in VAT collection. To start with, audit should focus on industries/sectors whose ratio of VAT-able purchases to taxable sales exhibit the greatest variance relative the I-O benchmark. But what is more worrisome is the lack of correspondence between the total negative VAT due in the previous year and the excess input tax carried over from the previous year in 1999-2001. This development suggests the absence of a good monitoring system for the excess input tax carried over from one period to the next and is clearly indicative of lapses in VAT administration. Finally, it should be stressed that, to be credible, industry benchmarking requires that each grouping represent a fairly homogeneous set of firms/ taxpayers in terms of key characteristics. Since VAT liability is largely dependent on the input-output structure of firms. In this regard, the 38 major sectors currently being used by the BIR in classifying firms appears to be inadequate. Prospectively, BIR should move towards a finer level of classification approaching at the minimum the I-O level of disaggregation.