This paper uses disaggregate inflation data spanning all of consumption to examine: (i) the persistence of disaggregate inflation relative to aggregate inflation; (ii) the distribution of persistence across consumption sectors; and (iii) whether persistence has changed. Assuming mean inflation to be unchanged, disaggregate persistence inflation is consistently below aggregate persistence. Taking into account an early 1990s shift in mean inflation identified by break tests yields much lower estimates of both aggregate and disaggregate persistence for 1984–2002. But with the mean break, average disaggregate persistence is actually as great as aggregate inflation persistence. A factor model provides a natural framework for interpreting the relationship between aggregate and disaggregate persistence. Copyright © 2006 John Wiley & Sons, Ltd.