I suspect that one of the two maps doesn't have enough points along the border to follow the desired projection's curve. To try to fix it, you could either use a less-curvy projection (such as cylindri), or you could leave more points in the map (by eliminating the where clause's "density<=something").   The longer explanation:    The reason it doesn't have enough points is (probably) that when the maps were processed by SAS (before  being shipped to the customers), Proc Greduce was probably run against the unprojected coordinates ... and the coordinates along that part of the US/Canada border pretty much have the same latitude, and therefore Proc Greduce thinks it doesn't need to keep very many data points along that border. But once you start using a projection, and those borders are shown as 'curves' then you need more coordinates to follow the curve.   One of the final projects I inherited and worked on before retiring was the processing of the maps. I changed the algorithm to project the map first, before running Proc Greduce, to try to reduce/eliminate this problem. I believe that the new maps created using my algorithm are the ones currently available for download (and are likely newer than the ones you have in your mapsgfk library). You might try downloading them, and see if it gives you better resolution along that border: https://support.sas.com/en/knowledge-base/maps-geocoding/maps-online.html   ... View more

In addition to what @ballardw pointed out, I would also suggest using the new maps from the mapsgfk library rather than the (possibly outdated) maps from the old maps library. One caveat there is that SAS only ships 1 Canada map in mapsgfk, therefore you'll have to derive the province map from it (using proc gremove) - that can be a little confusing if you haven't done it before, therefore I'm including some code below:   data map_us; set mapsgfk.us_states (where=(statecode not in ('AK' 'HI') and density<=4)); my_id=statecode; country='US'; run; data map_canada; set mapsgfk.canada (where=(id1 not in ('CA-60' 'CA-61' 'CA-62') and density<=1)); run; proc gremove data=map_canada out=map_canada; by id1; id id; run; data map_canada; set map_canada; my_id=id1; country='CA'; run; data mapall; length my_id $8; set map_us map_canada; run; proc sql noprint; create table mapall_attr as select unique country, my_id, 1 as fakedata from mapall; quit; run; proc gproject data=mapall out=mapall_proj latlong eastlong degrees project=gnomon; id country my_id; run; pattern1 v=solid color=white; proc gmap map=mapall_proj data=mapall_attr all; id my_id; choro fakedata / levels=1 nolegend coutline=gray55; run; ... View more

Thank you Dan and Kurt; I will speak with my SAS admin and have them make the needed changes. ... View more

No problem at all ...fastb. ... View more

Many thanks Dan; this is exactly what I needed to know! ... View more

Thank you Sanjay; as always, your solution worked perfectly! ... View more

Thank you so much Sanjay; that is the information that I needed! ... View more