Kinda suprised at responses here. Here's the medical field perspective.
BMI isn't a diagnosis. It's an indicator of potential problems only. So the medical field shouldn't use the values as the ultimate definition but that's not the point in BMI. It's there as an indicator of problem, not diagnosis of problems.
Through studies it's found that in the vast majority of cases it's actually correct, and especially so for low end results. The medical field is starting to catch on to the idea that it's more 27 before medical problems are an issue as opposed to 25 and this may be counting more for the discrepancies. In around 80% of cases where BMI was outside healthy the individuals were, in fact, not in the best health.
So if your BMI is below healthy, or above 27, should you see a doctor for a DIAGNOSIS of an indicator of issues? Yes!
BMI is like a thermometer reading. Just because you have a mild fever doesn't give you a diagnosis of definitely being ill, but it does indicate there's a potential problem which may need a professional diagnosis.