Any time a scientific method is used in evidence (whether in court or in family history or when-ever), it is important to consider its 'weaknesses'. Many researchers are not willing to do this, but I've always felt it is important.
DNA has three major weaknesses.
The first weakness of DNA testing is that all men in a direct relationship have the same DNA. For example, let's assume that several generations from now there is a Norton descendant who thinks he descends from Marc Norton. He could end up with a perfect match with another descendant, but he also would get a perfect match from a descendant of Ned, or Uncle Hub, or even a more distant 'Uncle'. DNA tells us that we are from a particular 'clan', but we still need to have historical documents to show us the exact relationships.
The second weakness of DNA is that once in a while there is a match with a family of a different surname. This could be random chance, since there are billions of people on this planet! More likely is that the two folks share common ancestors before surnames were ever used! Between 500-1000 years ago, surnames were first used by Europeans. It was even more recently than that in some cultures. There are obviously many people who share common ancestors from the pre-surname era, but have totally different surnames. I actually have a fairly good match with a man named Fishback (translated from German as Fish-Creek). His family and ours came from the same area of westcentral Germany at about the same time in the early 1700s. We almost certainly have a common ancestors. There is just no way to say who or when.
The third weakness of DNA is that sometimes there is no match when it seems that there should be. This is usually because of what is called a 'non-paternal' event. Some non-paternal events may have negative connotations. Examples of these would be rape or infidelity. Because we may never know the details of such a situation, or the reasons for such an occurrence, it may be better to leave it alone. Some examples of non-paternal events can be viewed in a much more positive light. Two examples of this would be adoption or step-family situations. If any of these situations are 'known' by future generations, there are no surprises, but as we all know 'family secrets' are very common, sometimes for good reason. It turns out that even if the odds of such an occurrence are low, when several generations are involved, it becomes quite likely that some sort of 'non-paternal' event may occur. If the historical documentation is good enough, even if the DNA does not match, it could be the correct family.
Since the Y-chromosome is what is tested in this type of analysis, every generation in the line of folks tested must be men. There are types of testing in which women can be involved, but because of the surname tradition in America, it is difficult to find a long line of women for whom the 'paper trail' exists. This means that there is no easy way to use DNA to verify our connection to families such as the Beasleys, Roes, Fosters, Woods, Hudsons. All of these lines are important to us, we just have to use traditional family history methods to verify the relationships.