End Game

End game.  It's one of the more difficult aspects of RPGs to get right.  A few possible reasons why:

1. Failure to Climax.  Most games don't reach the End Game.  They fizzle out along the way.  So GMs don't get a lot of practice ending a game. 
2. Lack of Planning.  How many times have you, as a GM, painstakingly planned the beginning of a campaign?  How many times have you planned the end game ahead of time?  Do you know what to plan for?  Probably not.  This is a often a problem with the sandbox game (see below).
3. Sandbox style almost works against an end game.  Because of the free world roam , do what you want style, its very difficult to plan far ahead.  But, sandbox does not mean there is shouldn't be a structure to the game and the events.  You may come up with a awesome conclusion and the players decide to go back to the tavern to stinky drunk and give all the wenches a bounce.  While not a bad way to end an evening maybe not the most exciting campaign exit.
4. Developing enemies over time.  When PCs encounter and enemy, they kill them.  Why release them so they can screw them over later?  You can build to a big boss, but there is something ultimately more satisfying having an enemy encounter through out the game before you drop the hammer on him.  
5. Campaign Scope is a rarely discussed topic and I think one that could help develop an End Game.  This is as easy as developing characters on a special mission that doesn't have to be of epic proportions.  Not every campaign has to be Lord of the Rings in epicness.  It's perfectly cool to have two slubs whose whole purpose is to infiltrate the enemy kingdom and steal away their secret weapon.  A finite series of events in time and space.  
6. Character Power gets to the point where the GM needs to throw ridiculous critters at them.  While the now 10th level party is constructing their Fortress of the Giant Cod Piece those bands of orcs and ogres and trolls are no threat.  Even the occasional dragon is a minor annoyance.  This is where a GM really needs to have his chops concerning politics and religion to make the game interesting.  But often times more effort is put into digging another level of a mega dungeon or stating out the beholder titan dragon god. 

Now are their more reasons?  Absolutely.  Are there ways to end a game with a party of 300th level characters in a sandbox style game with no arch enemy where the GM failed to plan it out?  I'm sure you can.  Everyone has their own style, strengths and weaknesses.  As a GM its important to have some sort of plan for your campaign.  Its important to have an overlying story going on that allows players the freedom, but are still effected by the events.  Without some sort of structure to the game the campaign soon degrades into a bunch of dudes looking at one another with pizza stained shirts and caffeine  headaches saying "What do you wanna do next?"