This paper investigates the relationship between aggregate consumption and permanent income using a new approach to the estimation of cointegrated systems that builds on Stock and Watson's common stochastic trends representation. The permanent and transitory movements in aggregate income and consumption are estimated directly using the Kalman filter and are allowed to be correlated. This approach avoids any implicit restriction that permanent income be as smooth as consumption. Instead, permanent income appears to be relatively volatile, with consumption adjusting toward it only slowly over time. These results provide a clear rejection of the standard version of the permanent income hypothesis and are suggestive of alternative theories of consumption behavior such as habit formation or precautionary savings.