An analysis is presented of physical phenomena occurring during secondary-side feed-and-bleed, which is being considered as an accident management procedure for pressurized water reactors. Problem areas related to the reliability of this procedure are identified and discussed. Secondary-side feed-and-bleed, as examined in the LOBI (which is the Light Water Reactor Off-Normal Behavior Investigation) integral system test facility, was not successful because of a delay in the secondary-side heat removal caused by the release of stored heat from the steam generator downcomer wall. Countercurrent flow limiting in the hot leg was another major phenomenon that can influence the effectiveness of this procedure. The discussion of the experimental results is complemented by relevant calculations byRELAP5/MOD3. In general, it was found that the effectiveness of the feed-and-bleed procedure is maximized if the number of steam generators used to implement it is reduced.