It depends. We are learning about this right now in developmental psychopathology.
I dont believe that each subsequent DSM has made it worse for children. In fact, comparing the previous DSMs, the current DSM (DSM-IV-TR) has actually been very good for children and adolescent diagnoses.
I dont believe that diagnosing children and adolescents via the DSM will ever be as accurate and reliable as diagnosing adults. That is one of the big issues with the DSM. We keep getting more reliable and accurate with diagnosing adults and we are trying to play catch-up with diagnosing children.
One of the bigger problems is the medical model used for creating and diagnosing psychological disorders. While it is a great base for it, it is not the end all and some professional still use it as the only way to diagnose children.
Luckily, that seems like it is changing because we are learning less and less about the medical model and more about the multiaxial system and using not as mnay concrete methods. The problem is, to be covered by insurance companies and to be used by professionals, they have to be empirically based so there are limits as to what can be done. There has to be some concrete evidence to test for reliabiity and validity of the methods.
Obviously, this is not an opinion coming from someone who has a child that has been diagnosed. This is just from what I have expereinced by going for my psychology degree and what we have been learning so far.