How Long do you Try Repairing Before Reformatting?

As a small-time computer consultant I've dealt with dozens of customers who bring me computers loaded with viruses and spyware.  I'm of the school of thought that a reformat is never neccessary, that--given enough work--any infestation can be overcome.  But there is a point of diminishing returns, and I'd like to know where others think that point lies.

I've arbitrarily decided that at my current rates it's no longer worth my customer's hard-earned money to try and root out an infestation beyond two hours of labor. After that I suggest a reformat.  Where do you draw the line?