Class distinctions are about significant qualitative distinctions, not points on a simple quantitative distribution. A reasonable upper/middle/lower class distinction could be made where the upper class has capital assets that they can and do live off the proceeds of while increasing the value and annual proceeds of the assets, the middle class has capital assets that they could (whether or not they do -- they may still work at wage labor) live for at least some months on by depleting, and the lower classes are entirely dependent on wage labor (and one could make further distinctions within these groups).
This would roughly correspond with the traditional ideas of "class".
EDIT: Which is all not to discount the potential significance of an ongoing increase in income inequality -- which almost certainly has some real impact in terms of the kind of things that would reasonably define class.