Throughout most of the democratic world, elections results tend to split around regional differences. Historically this is true of the United States as well. Think back to the American Civil War. In presidential election of 1860 that preceded the war, the country split almost perfectly along a north south divide. The following map illustrates the stark divide in politics between north and south.
However, over the last 20-30 years in the United States, this long-standing pattern has begun to unravel. With each passing year, the dividing line between Democrats and Republicans has shifted more and more from a regional separation to one of urban vs. rural. Cities have increasingly swung Democratic while rural areas have become Republican strongholds. It seems that population density is becoming one of the biggest indicators of voting preference.
Take of example this map of midterm election results for 2010. Notice that most of the country looks strongly red; the Republican party controls a strong majority of the physical land area.
Notice also that this map is very different from the first. There seems to be few regional splits in voting preferences. Blue and red areas are largely intermixed throughout the country.
But now look at this third map. This shows the same results as the last map, but now the sizes of the districts have been changed to show them in proportion to their population. This image is barely recognizable as the United States. Notice how tiny the huge swath of red districts has become in comparison to the now prominent blue, Democrat controlled districts. You can clearly see Cape Cod, Long Island and the San Francisco Bay Area all colored blue.
These blue areas almost exclusively represent densely populated urban areas. When looking at this map, the country looks much more evenly split. In the last election in 2012, 27 of the 30 largest urban areas in the country voted Democratic in the presidential election.
If this trend continues with urbanization, the future of the Republican party as we know it may be uncertain.