That is correct. Differently sized bills enable blind users to differentiate them without having to rely upon the honesty of a cashier or a complicated filing scheme (and, additionally, allow sighted people to more-easily differentiate denominations in low-light conditions).

Edit: specifically, I learned about this from Gordon Legge[1], the chair of the psychology department at the University of Minnesota when I took a graduate level course on human-machine interaction from him back in 2002 or 2003. I lucked out and had a very forward-thinking advisor in the Computer Science department who let me apply Professor Legge's class to my Senior year electives ;)

[1] see http://en.wikipedia.org/wiki/Gordon_Legge and http://vision.psych.umn.edu/users/legge/ for more on him.