Traditional or conventional working dogs is a term I use to lump together dogs like GSD, Mal, Rott, labrador retriever etc. These are working dogs that were designed in one capacity or another to work closely with the handler. I find that the temperament tests I know predict fairly accurately how the dog will turn out. One of the reasons they can turn out the way you expect is because of the trainability these breeds possess. Once well selected a puppy through correct upbringing can be molded. These dogs have a degree of malleability and once the conditioning is set it holds at some of the deepest levels of temperament. By contrast I consider molossers to have been developed to work independently. While some possess a good degree of willingness I still find trainability low. With these dogs I think what you got from the beginning is what you get so to speak. Training doesn't hold the way it does with the traditional working breeds. They seem to always be thinking about what to do when you require something of them. Because of this it seems to me that puppy temperament tests may not hold in these dogs because you can't follow through with molding the dog other than some of the most basic things like house breaking for example. These have been my observations so far as I have a lot to compare the molossers to. I've never considered temperament tests as a way to avoid mismatches. I always looked at temperament tests as a way for me to get the dog I want which must possess certain temperament qualities necessary for a protection dog. As mentioned before more dogs are not successful in their homes because most people are idiots when it comes to dogs. I don't think there's another field in existence where so much BS is accepted as fact and everyone's an expert. [blockquote]Think about the high attrition rate of seeing eye dogs in training or explosive detection dogs.[/blockquote] I disagree with this statement. If you read the New Knowledge of Dog Behavior by Clarence Pfaffenberger you will see that they got the success rate of seeing eye dogs to something like %90 once they tuned their breeding program and puppy selection process. Same goes for narcotic detection dogs. I'm familiar with the training of narcotic detection dogs but not with explosive detection dogs. May be the same process except that I believe for explosives a passive alert is required.