This paper investigates the econometric properties of the Taylor (1993) rule applied to US, Australian and Swedish data to judge its empirical relevance. Unit root tests indicate that the variables in the Taylor rule are near integrated processes, implying that cointegration is a necessary condition both for consistent estimation of the parameters of the model and compatibility between the model and the data. Tests find little support for cointegration and, together with an out-of-sample forecast exercise, suggest that we should have serious doubts about the Taylor rule as a reasonable description of how monetary policy is presently conducted. Parameters in Taylor rule regressions are therefore likely to be inconsistently estimated, and caution should be taken before central bank policy is evaluated using such methods.