... in some fields, there is a lot of traffic on the arXiv, and people watch their subject's section to see what's new and who's publishing it.
Especially if you're young, I'd think that it's a good idea to publish things in peer-reviewed journals, so that those people who are evaluating you can check that others in your field value your work. If you never publish in peer-reviewed journals, then people might start asking questions about the quality and/or sanity of your work.
The way that I rank journals in my area is to ask my colleagues (I'm a number theorist; Inventiones and the Annals of Mathematics are both good journals, for instance). You could also look at some back issues to see what sort of things appear and who wrote them.
One bad sign is when papers in subject X appear in journals that don't normally carry anything related to subject X. This may mean that the peer-reviewing has not been of the highest quality; the journal's normal reviewers may not have been able to judge the paper's merits.
[ Parent ]
