My experience - size doesn't matter
Okay - what really makes a difference is the std used in the curve -
One wants that std to be as similiar as the protein you are looking for.
For example, using BSA and testing for a plant protein doesn't work very well.
Let me recap
1) I've never seen the size of proteins to cause a problem w/Bradford
2) what one uses for the std does matter. I've seen lots of problems with this. It does affect the accuracy of the test.
3) at higher conc the std curve doesn't alway remain linear
4) watch out for impurities like detergents
that's all I can think of right now