Currently, there seems to be little official definition, and the term MAV appears is relatively new. The only paper on the topic I could find that specifically approached the area defined it as:
"small enough to be practical for a single-person transport and use." (Galiński and
R. Żbikowski, 2007)
Delft University of Technology section on MAV research provide no clear definition in their documents.
I think the definition seems to more subjective to the flying properties to the UAV, which might make the definition more difficult. Since they are not carrying people but sensors, they will be much more varying in weight, especially in with time.
One example, the british army's MAV has a flight time of 25 minutes. This one is a very small but has a respectable flight time. This will not simplify definition. I think the term coined by Galiński and Żbikowski is the most useful in a practical sense.
Update:
University of Texas page on the DARPA competion says:
The definition employed in DARPAs program limits these aircraft to a
size less than 15 centimeters (6 in.) in length, width, or height.
MAVs must have a weight of 50 grams or less and must be capable of
staying aloft for 20 to 60 minutes for a distance of 10 kilometers.
The pure source for this this, while not avaliable any longer, I believe I found to be
The definition employed in DARPA's program limits these craft to a
size less than 15 cm (about 6 inches) in length, width or height. Source
There are currently no civilian classifications for UAVs from the FAA or EASA, hence the legal problem with these. Most army and air force drones operate under military exceptions if I'm not mistaken.