We propose and apply a new approach for analyzing the effects of fiscal policy using vector autoregressions. Unlike most of the previous literature this approach does not require that the contemporaneous reaction of some variables to fiscal policy shocks be set to zero or need additional information, such as the timing of wars, in order to identify fiscal policy shocks. The paper's method is a purely vector autoregressive approach which can be universally applied. The approach also has the advantages that it is able to model the effects of announcements of future changes in fiscal policy and that it is able to distinguish between the changes in fiscal variables caused by fiscal policy shocks and those caused by business cycle and monetary policy shocks. We apply the method to US quarterly data from 1955-2000 and obtain interesting results. Our key finding is that the best fiscal policy to stimulate the economy is a deficit-financed tax cut and that the long term costs of fiscal expansion through government spending are probably greater than the short term gains.