In all seriousness, this was fascinating to me. I appreciate the lenghty explanation of both Ennay and gObLuE (I read it twice) and you've flipped me. Well done.

Lank, you're so damn spineless. What a flip-flopper. Sticking with a position, even in the face of insurmountable evidence to the contrary, is courageous, admirable and is always far superior to someone who changes their mind.

Lank, you must be like Stephen Colbert, who said, " I'm not a fan of facts. You see, the facts can change, but my opinion will never change, no matter what the facts are."

Maybe you should say it here. Or maybe not, depending on how you roll...

Lank, you're so damn spineless. What a flip-flopper. Sticking with a position, even in the face of insurmountable evidence to the contrary, is courageous, admirable and is always far superior to someone who changes their mind.

That's funny. Wait. No it isn't. Wait. Is it? I need someone to make up my mind because I can't decide.

To answer the OP's question, if a 5k course is certified, you have reasonable expectation of 5000 + 9/-0 meters, based on Ennay's numbers. If it's not certified, than it I'd day that the ±2% number that's been floated out there seems reasonable. But on an uncertified course, it depends a lot on who's setting the course up. A small race with the course laid out via a car odometer -- not so good. But a race with participation from a running store, running club, or some organization will most likely be fairly accurate.

The certification process is about the best we have and it's usually darn close. Last month I had a talk with a long-time USATF course certifier, and let me tell you--these guys are very anal about the accurancy of these courses, to put it mildly. It amazes me that so many many people will question the accuracy of a certified course because "my Garmin said...". A Garmin measurement isn't nearly as reliable as a certified course. If you want be sure that you actually ran as fast (or slow) as your watch said, race on a track or a certified course. But since it's not always possible to find certified courses and there are some pretty nice races that aren't certified, here is a method for the obsessive, "type A" nuts like me: Find the results of the previous running of the uncertified race you are interested in. Next, take a sample from the first 20 finishers and see what times they ran on other known certified courses. You can use a local results site, or one of the big ones such as athlinks.com. Compare those results with the uncertified ones and see if they line up closesly. If they do then it's probably safe to assume that the uncertified route is accurate.

Also, as far as a 2% error for uncertified courses, this is bullocks. Basic statistics, folks. Here is the deal: All courses are either the correct distance or they are not. Courses that are not the correct distance have an error. There is no reason to believe that a course with an error has a higher degree of error just because it was or was not certified. A statement that is more likely correct is this: There is a 1% chance that a certified course is the incorrect distance while there is a 5% chance that an uncertified course is the incorrect distance. (these are, of course, funny numbers to provide an example) Also, please remember that there are several steps to getting the correct distance - - measurement, in which the course is measured by a standard method* - layout, in which the race director and team sets up the course correctly along the measured course - race, in which the runners actually follow the measured route along the measurement line. *this is where certification comes in

GPS error

This is one of the best GPS accuracy references I've seen yet. Excellent information.