Thanks for the additional insight. The testing with model builder was a good call. The code generated from that (that worked fine) below.It is still somewhat unclear to me why my original code doesn't honor the split though. I don't believe the shapefile is the issue since I am creating a feature layer using MakeFeatureLayer_management. And, when I check the the field_info properties using getsplitrule after creation I see, for my numeric field in question, that "RATIO" is listed.I do appreciate the explicit connection between the created layer and the split policy that is evidenced in the model builder code version though.As to Wayne's point about apportioning population: I'm with you about the population density issues. However in this case this is census block data and the spatial resolution is fine enough where we're not worrying about minor spurious re-apportionment. It is actually useful to have sliver polygons that are output from the intersection have populations of 0 or 1 so we can catch the error easier.import arcpy
# Local variables:
blocks_01_alb_shp = "...\\blocks_01_alb.shp"
USA_int_131111 = "USA_int_131111"
State_01_int_7_shp = "...\\State_01_int_7.shp"
blocks_01_alb_Layer1 = "blocks_01_alb_Layer1"
# Process: Make Feature Layer
arcpy.MakeFeatureLayer_management(blocks_01_alb_shp, blocks_01_alb_Layer1, "", "", "FID FID VISIBLE NONE;Shape Shape VISIBLE NONE;STATEFP10 STATEFP10 VISIBLE NONE;POP10 POP10 VISIBLE RATIO")
# Process: Intersect
arcpy.Intersect_analysis("blocks_01_alb_Layer1 #;USA_int_131111 #", State_01_int_7_shp, "ALL", "", "INPUT")