Immunoblot semi-quantitative at best?

[Jeff Brown]

I have read that Western blotting is best used to determine the presence 
or lack of presence of a protein, and should not be used as a 
quantitative tool.  The reasoning is that the antibody-antigen 
relationship and antibody-antibody interactions are not a one-to-one 
ratio. Also, homologous antibodies are not always available.

Is this the consensus? Is ELISA a better quantitative tool?

Thank you
Jeff