I looked closely at Ajay's image. In the upper right portion, there is a stream that goes outside of the watershed boundary. This is likely the problem. See the attached close up of the spot the creek leaves the boundary. If you modify your boundary to include the stream, things should get better. All your grid data (fdr, rawdem, str, strlnk, etc...) need to be inside the boundary also.
I once had a creek that went right up to the edge of a watershed. In the processing creek would end right near the boundary. I think it is because the burning streams lowered the DEM at the boundary and then the flow direction grid naturally choose that as the outlet. The solution was to use an "outer wall". This is likely what is needed here.
One problem could be with the "polygon file of [your] actual watershed"(1). In the hydrology geoprocessing, nothing is the "actual watershed". The DEM is a discrete grid. If you look close at the streams generated, they go from grid center to grid center. The watershed boundary's do the same. This does not follow the "actual" feature (stream or ridge). So, if you try to superimpose an "actual watershed" on the grid, you need to consider that the polygon you bring in may not line up just right with the DEM.
[ATTACH=CONFIG]26070[/ATTACH]
(1)This assumes that I understand what your doing with the "actual watershed". Did you load your "actual watershed" just for comparisons, or did you load it to use it in the processing (Batch Watershed Delineation for Watersheds)?
