Are you seeing a lot of route changes where the route length changes? This is a very difficult problem - that is getting just the right amount of aggressive vs conservative in dispatching packets so we don't double-query and count for a specific target while still making sure it gets requests.

If the route is constantly lengthening and shortening, trace route is sometimes not very reliable (you might get similar results with the TRACERT command). We've addressed this in an upcoming release by doing a hybrid technique for the final destination that will sometimes not catch every route change exactly right, but it will get the latency and packet loss correct on the final destination. If you're a licensed user, you can grab that from http://www.pingplotter.com/beta and see if it solves your problem.