Abstract Principal component analysis (PCA) is applied to six macroeconomic time series observed over 1959–2007. Six periods in US economic history are identified by a cluster analysis of observations in the PCA score plot. The method is data driven with no a priori information on the number or dates of breaks. Our findings give independent support to the effect of the oil price shock in 1973, and the introduction of the Great Moderation period. Of the five transition periods, two have been identified by previous studies as breaks (1973, 1984), one is a well-known date of monetary policy change (1979), and two had not previously been identified (1970, 1977–1978). In the long-run inflation and the federal funds rate are unrelated to industrial production and unemployment. Inflation and interest are positively associated as predicted by the Fisher hypothesis. These long-run relations argue against the use of monetary policy to peg the rate of unemployment or real interest rates. In the short-run inflation acts a leading indicator for unemployment for the period 1959–1997, but not for the period after 1997. The well-established reduction in macroeconomic volatility in the mid-1980s is specific to the period from 1985 to 1997; volatility subsequently rises above pre-1979 levels.