This paper presents a test of the Friedman hypothesis: Friedman (1977) argues that increases in the average inflation rate are often associated with a rise in inflation variability and hence inflation uncertainty. With reference to the importance of the time horizon in analysing inflation uncertainty we utilise an unobserved components model of inflation which decomposes inflation uncertainty into two measures, one short term, the other long term. Results obtained from a panel of data for the G7 countries provide support for Friedman's basic contention that inflation uncertainty affects real output. In particular, long-term inflation uncertainty has a negative effect on real output. Our results also underscore the importance of central bank independence as a possible influence for fluctuations in real activity.