If I have a paper accepted for publication in a journal, putting a version of it (that does not conflict with the agreement with the publisher) on arXiv will help make it more easily accessible. On the other hand, are there good reasons not to put it on arXiv, supposing the paper is on topic for arXiv?
It's really hard to find cons. If I were forced to list some, I'd say:
people get to see your LaTeX source. If it contains secret macros or unflattering comments that you forgot to delete, the whole world sees them. It is also possible that they contain other copyrighted material that you do not have the right to distribute, such as a commercial font. (In my experience, no one will look at your LaTeX source anyway).
Arxiv gets rights to distribute your paper online, forever. This may conflict with copyright rules for your future publications. For instance, you may want to insert content from the paper into your thesis, and then publish your thesis as a book with a publishing house that wants full 100% absolute exclusive rights. (In my experience, respectable publishing houses don't.)
Your publication agreements or grants may mandate additional requirements when you publish your work. For instance: provide a reference to the journal version, or acknowledge the National Institute of Funding Grant 1234567890. If you forget about them, trouble may ensue. (In my experience, it hardly ever does).