BMI is a measure of fatness.  
We the measures of fatness overweight, obese and morbidly obese and use numbers from BMI- body mass index- to categorize people.  We are essentially referring to risk associated with percent body fatness, not weight.  BMI is a proxy measure of fatness - excess adipose tissue.  This excess relates to risk for disease. The BMI is helpful for researchers more than individuals because when you have 100s and 1000s of people with BMIs at a certain level grouped together, MOST of them will meet the criteria for overfat by their BMI number and those who do not won't really change the analysis.

My BMI does not declare my health status, but if I had a high one, over 25, I would want to ask my doctor if my fat levels were of  concern.  This is an individual issue.

The risk for disease in an "overweight" person is not as great as that in an "obese" or "morbidly obese person."  Generally speaking.  If I am overweight I may have a certain risk of high blood pressure, let us just say 10%, but if I am obese, that rises to 20%.  My BMI doesn't really tell me enough though- even if I look at a chart or use a calculator to determine it.  I might find myself in the >25<30 cat or the >30 cat or the >40.  Certainly - all of us are at great risk when it is 30 or higher, but those 20s can be a little more nuanced.

This difference is not always small enough to be ignored in data analysis either, I am learning.
In reviewing numerous articles within the topic of obesity and chronic disease, I have started to see that the amount of fatness (better measured with waist circumference, skin calipers or other techniques beyond BMI) that increases risk for disease is 1) not the same for each disease (e.g. heart disease or diabetes) and 2) is not the same overall or specifically by race or ethnicity.

I will have to read a lot more to fully grasp the numbers and explanations (some of them as yet unknown) but I can say a few words now and keep this issue in mind.  [It seems the more precise our measuring, the less generalizations we can make!]

For example, current wisdom is that a BMI over 25 is an indicator for disease risk in a population of people (as a group).  It turns out that Asian Americans may have disease risks with a BMI at 24 and black Americans are fine up to 26.  Thus, the number of whites in the samples used to determine these things may have skewed the numbers.  At the same time, a white American may have a higher risk of heart disease at BMI 25 than a black American, but a black American a higher diabetes risk at a lower BMI.  Gender differences are there as well and these numbers were just for demonstration purposes!

I suppose the 'bottom line take home' points are these:
Body fat, especially visceral fat, is linked to disease.  One way that we try to measure it is BMI but it is very inaccurate at the individual level.  For an individual, WC or WHR ratio might be better. You can click here to review that information.
Most of the time, if you have extra body fat, you know it. :)  You may not want to know it - because then you have to do something about it - but you know. 
Lastly, we each have our own tipping point where our weight makes us sick and it can be higher or lower depending on many things. Additionally, we have a physical and psychological weight preference, but it is less important with regard to disease causation.