All depends on how accurate you need your course to be. I've used measuring wheels as well as consumer GPS units. I also measure courses for certification using the Jones Counter. The advantage the Jones Counter has over other methods is that the outcomes are easily repeatable, within very small deviations. One 5k course I helped measure in 10/2005 I went back to remeasure in 11/2015 when our 10 year certification expired and my measurements were within 6" of what we got in 2005.
With any of these devices, the experience of the measurer goes a long way towards getting results with decent accuracy. If you're new to it expect to make some measurements that you go back to later and remeasure and get slightly different results - moreso with the GPS and measuring wheel than with the Jones Counter method
Measuring wheel - I frequently use one to measure things for the XC team I coach. Measuring off pavement produces much more variance than on pavement. We have a loop through the woods that I measure every few years to see if the corners are changing and I've learned I'll get as much as a 10' variance (+ or -) over about 1500m. On the other hand, I have tested a measuring wheel against my Jones Counter measurements on some local road 5k courses and I'm usually within 3' over 5k. The wheel always shows longer for a given segment.
For GPS I use a Garmin eTrex30. I use it a lot hiking in the woods and it consistently loses .2-.3 or more over a 10 mile hike. For example, I hiked Sandwich in NH this fall. The White Mountain Guide says the loop I did was 8.2 miles, the eTrex read 8.04 back at the trailhead. I haven't used the eTrex much on runs, but I did bring it along when I did a training run on a certified 10k course this fall and got 6.16 miles instead of the expected 6.21. That course is pretty wide open, just a little tree cover at the start and finish. It's mostly quiet dirt roads so I was to run tangents almost everywhere. In a normal race situation, not being able to run the tangents as religiously, I'd expect the eTrex would have given me something in the 6.3 range on this course (not a lot of turns). Personal GPS is always going to give you a longer read in a typical race situation when you're not running the exact tangents perfectly.
For an experienced measurer I'd say a wheel is marginally more accurate than current GPS in most situations. But the GPS should give you entirely useful measurements and it's a lot quicker. Sort of depends on the situation. If I'm measuring a loop in the woods with thick tree cover, as long as it's not a long loop (a couple miles max) I'd probably use the wheel. If it's an open field I'd jog it with the eTrex and call it good.
