The International Comparisons Program (ICP) run by the World Bank compares prices and real incomes across countries, and plays a pivotal role in the Penn World Table. Using a unique dataset consisting of over 600,000 price quotes from nine countries in the Asia‐Pacific region, we consider ways of improving the basic heading price indexes that form the building blocks of ICP. Current ICP methodology computes these price indexes using the country–product–dummy (CPD) method applied to the country average prices. We contrast this approach with: (i) a weighted version of CPD; (ii) CPD applied directly to the individual price quotes; and (iii) extended versions of CPD that include adjustments for unrepresentative products, urban–rural price differences, and different outlet‐types. Also considered are new CPD‐based methods for measuring urban–rural price differences, and the implications of our findings for the downward revision in China's GDP in ICP 2005.