Abstract This work examines the life cycle of sudden stratospheric warmings (SSWs) from composites of a large number of events. The events are sampled from idealized general circulation model (GCM) integrations and form a database of several hundred major, displacement, splitting, and weak vortex events. It is shown that except for a few details, the generic zonal-mean evolution does not depend on the definition used to detect SSWs. In all cases, the composites show the stratosphere in a positive annular mode phase prior to the events and a barotropic response in the stratosphere at onset. There is a clear positive peak in upward Eliassen–Palm (EP) flux prior to the onset date in the stratosphere and a much weaker peak in the troposphere, making the evolution more consistent with the picture of the stratosphere acting as a variable filter of tropospheric EP flux, rather than SSWs being forced by a strong “burst” in the troposphere. When comparing composites of SSWs from the database with apparent influence at the surface (downward “propagating”) to those without such influence, the only significant differences are a somewhat more barotropic response at the onset date and longer persistence in the lower stratosphere after the onset for propagating SSWs. There is no significant difference in EP flux between propagating and nonpropagating events, and none of the definitions considered here shows a particular skill in selecting propagating events.