There is a lot of controversy over check dams in watershed restoration circles. The folks at Quivira Coalition, as well as Neil Bertrando and
Brad Lancaster, have spoken out against check dams in favor of boulder weirds, one-rock dams, and other tools that you can read about in "Let the Water Do The Work". The main argument against check dams is that the seed of their destruction is a result of their main function; eventually the rocks in the check dam erode away from the trickling water and the whole thing collapses in a wash.
In my own work, I have built hundreds of check dams and rather than relying on the structural stability of the stone, intend to use taprooted and shallow-rooted plants to stabilize the structures as they retain more and more water. I think with plants stabilizing the hydrology and slowing down erosion, and by using large boulders instead of small stones, we can avoid the pitfalls of check dams and create structures that will last a very very long time.
Check dams are not for the untrained to use though; they're tremendously easy to do poorly, but poorly done they almost never function well. Furthermore, I don't think they
should be used at all in eroded gullies, but in more stony and mountainous regions where erosion is less likely to undercut them.