Cross-national price comparisons for pharmaceuticals are commonly used for two purposes. Comparisons based on a sample of products are used to draw conclusions about differences in average price levels. Cross-national comparisons applied to individual products are also used by governments to set domestic prices. This paper examines the major methodological issues raised by international price comparisons, focusing on measurement of differences in average price levels and the validity of policy conclusions drawn from such price comparison studies. It argues that valid measures of average price levels can only be obtained from comparisons based on a comprehensive or representative sample of products, appropriately weighted, following standard index number methods. Comparisons of individual product prices should take into account the manufacturer's entire product portfolio over time rather than focus narrowly on a single product at a point in time. Because of the great variation across countries in both the range of drug compounds available and the dosage forms, strengths and pack sizes for each compound, obtaining a broadly comprehensive or representative sample is problematical. If products are required to match on all dimensions, including molecule, manufacturer, strength and pack, as is common in most international price comparisons, then only a very small and unrepresentative sample of the drugs available in each country can be included in the analysis. A trade-off between the desire to compare only identical products and the need to compare a truly representative sample of a country's pharmaceutical market is therefore necessary. A valid comparison of average drug prices should include generics and over-the-counter products that are good substitutes for branded prescription drugs, with all forms, strengths and packs. To achieve this broad representation, however, the requirements of same manufacturer, same brand, dosage form, strength and pack size must be dropped. When such an approach is taken to the comparison of international drug prices, quite different results from those obtained from less comprehensive comparisons may be obtained. Indeed, a major conclusion of this analysis is that international drug price comparisons are extremely sensitive to choices made about certain key methodological issues, such as sample selection, unit of measurement for price and volume, the relative weight given to consumption patterns in the countries being compared, and the use of exchange rates or purchasing power parities for currency conversion. In particular, the results of this analysis indicate that recent reports suggesting that manufacturer prices in the US are 32% higher than in Canada and 60% higher than in the UK are in fact overstatements which arise from limitations of the sample and methods used to calculate these price differentials.