Define Elevation Gain for me...

This is correct.

I wouldn't necessarily trust gmaps or google to give an accurate elevation reading.  Need something that records barometric changes like the Garmin Edge (even the Forerunner, which uses GPS, is notoriously liberal in giving total elevation readings)

I've noticed that sometimes with the route mapper feature that the elevation changes get goofed up.  What I discovered is that, if you open up the course in edit mode, there is a button to 'Reset Elevations' or something like that.  If you click that, at least in my experience, it comes up with a much more reasonable number... 

Correct, elevation gain is the sum total of all climbing done over the course as determined by the points created by clicking on the route. So to get a truly accurate elevation gain value you would need to click at each point where elevation changes.

We don't get elevation values from Google. The primary source for this data is a web service run by the U.S. Geological Survey and if that's down, we use GeoNames which uses a less precise dataset and is a bit slower, but it's better than nothing.

If you have a route that has a flaky elevation gain value, edit the route and use the Reset Elevation Values button. I used it on the route linked to above in Dallas and got a more reasonable result of 335 ft.

Here's a tip: The new Terrain View on the Route Tracker can be very helpful for this. It shows elevation changes quite well. More gradual changes are seen better with wider zooms and smaller changes with tighter zooms. When creating a route, use Terrain View and create points as near the tops and bottoms of hills and valleys as you can. Points can be dragged if needed.

Also, total climb values on the routes I've been test uploading directly from my Garmin 305 to the Route Tracker have generally varied only a bit (±2%) from the same routes that I've plotted out manually.

Wait, did I say that out loud?

Sounds like you received some good answers to your inquiry about elevation gain.  Enjoy your 50K.  Is it your first?

Apologies for continuing with the Garmin 305 hijack.  Regarding the Garmin 305, I use the SporTracks software.  One option in the software is to set the smoothing interval.  Because of the inherent inaccuracy of the GPS (I think it's 15' at best with the max number of satellites), the smoothing function, well, smooths out all the reading by reading fluctuations.  I generally use 45 seconds, but in reality, I don't have a good reference to know what the correct amount of smoothing is.  Anyone else have any thoughts on that?

This is right on - but to add a little more - I trust the barometric pressure altimeters a little more than GPS.  As you may or may not know, a GPS triangulates your location.  So, it takes a reading and may put you on the planet 30' from where you stand.  The next second, it puts you 30' on the other side of where you stand.  It takes the averages and assumes you are in the middle.  It is constantly taking readings and adjusting your position.  The more satalites you get, the more accurate the reading.  But since your position is always an "average" of the shots, you would appear to move even if you are standing still.  This error is even bigger for your vertical location.  In fact, if you just stand in a location for an hour with GPS, it will tell you that you had several feet of assent and descent after the hour.  As for the BP device, I think it is a little more accurate, but atmospheric conditions/changing pressure systems can make it look like you assend/descend as well.