Chaining is used in index number construction to update weights and link new items into an index. However, chained indexes can suffer from, sometimes substantial, drift. The Consumer Price Index Manual (ILO, 2004) recommends the use of dissimilarity indexes to determine when chaining is appropriate. This study provides the first empirical application of dissimilarity indexes in this context. We find that dissimilarity indexes do not appear to be sufficient to resolve the issue of when to chain.