A lot depends on details of history too, such as the pattern whereby civilization comes to the continent. Are we imagining an ATL Native analog to the Roman Empire somewhere (on the Great Lakes? The Gulf of Mexico? The Atlantic Coast? The Southwest?) and assuming some ATL domestication of some draft, meat, and/or fur-bearing animals to allow civilization to follow a more Old World sort of pattern, so that states on a Roman or Han Chinese scale are possible? Does the High Middle Ages equivalent American society arise over a region that is partially a portion of the old empire and partially territory that empire never held? Where are the ancient antecedents to the "medieval" societies--does civilization largely radiate out of MesoAmerica as OTL, or is there a very different cradle such as the Great Lakes or some part of the Mississippi Valley or even say Chesapeake Bay or somewhere on the West Coast--California, Columbia River, Puget Sound?
Who comes first, what they have and where they have it will make a big difference. Of course if economics bypasses a once-great city that city will wither and quite likely vanish from the map. But we may well have two sites, essentially equal in merit, where one is highly developed and has been for several thousand years, and the other is a recent upstart, simply due to the pattern whereby civilization arose and spread.
If we aren't inventing ATL ancient domestications that totally transform the Old World, then a medieval North America can't really come into being any later than the European Middle Ages. Either some Old World ancient civilization branches into the New World somehow--for Romans to be able to do this, settling in the Caribbean let us say, has been a TL subject but I never was persuaded they could do that. It would probably be easier for East Asians from China, Japan, or Korea to arrive in Classical times, but still very very difficult and it is hard to see what would motivate an expedition to say California, let alone substantial settlement there. An ancient Old World civ somehow establishing itself in say Puget Sound (coming from Asia) or the Caribbean (coming from the Roman Empire) around the beginning of the Common Era might grow, spread and mutate into the foundation of a broad "medieval" zone a thousand years hence--though it would probably lack the constant injection of new tech and ideas that helped shape the sequence of European history.
Alternatively we could be talking about a colonization of North America during medieval times, with the colonies being not much behind their mother countries in technology.
In either of these latter cases, the direction of settlement will have a lot to do with the actual pattern of urban development.
All that said, I suppose that on the whole, while this determines the general level of development of large regions--which are old settlements developed to the hilt already and which are new colonies being carved out of wilderness just in the past generation or so--the exact sites of the greater cities of the region, whether these are ancient city-states or brand new outposts, will be mostly the same as OTL prosperous towns that have some other cause than their position on the right rail lines for glory. And by and large, most really major American cities have something going for them other than some arbitrary historical choice. Even when it was a RR that fostered a town, chances are that particular spot on the line had something special about it, and the rail line ran through that point for reasons that would tend to channel other modes of transport near it. There's a pass it leads to, or a river brings reliable water in a scrub land, or a lake makes an oasis in the desert.
There must be some interesting exceptions, some fair-sized modern American towns that simply would never exist under medieval conditions, or vice versa would be mighty in medieval conditions but OTL are bypassed.
---
Perhaps, you the OP who has presumably thought about this problem a while, might sketch up a map that highlights sites suitable to base a substantial medieval presence where moderns would not particularly choose to site a city? The criteria governing the location of a castle might be quite different from those governing where to put a town. (I'd think though that even in medieval conditions, a castle might be put where a town is uneconomic, but another one that is put where a town would prosper is more liable to become a major, important, retained and rebuilt over the centuries rather than abandoned castle!
Vice versa, starting with a map of OTL one might eliminate, or anyway downgrade, towns in places where we'd expect a lot less civilization than in modern conditions, such as the above mentioned Great Plains that medieval people could not farm, or places in the middle of deserts--but be careful with that, since current desert towns tend to b on the major transport routes, and these are close to the old, pre-railroad trails for crossing these. So odds are an AT North America that has been medievalized somehow or other would have some sort of settlement along those routes after all--but it would be constrained by the available water. Some spots have water supplies that aren't evident to the naked eye; if these are simple enough to access those towns can be quite large, if we are talking supplies that can irrigate fields with medieval tech. Other spots would be reduced very severely indeed.