Publication in proceedings Why and how publishing a paper in proceedings?
What are the difference with a "classical" journal?
What's the list of the main proceedings in which one can publish?
Do proceedings papers (never, sometimes, often or always) appear on mathscinet?
 A: I agree completely with Andreas' answer.  One further consideration is publicity.  It is easy for papers published in conference proceedings to become lost to general knowledge, or known only to very specialized groups.  By publishing in a regular and reputable journal, the chances others will read your paper goes up.
Further, it is not only administrators who hold the opinion that many conference proceeding volumes are of lower quality (at least in mathematics, but not, say, in computer science).
A: Proceedings of conferences are often published as special issues of "classical" journals. But even those that are not are usually included in MathSciNet if they include a statement (often a footnote on the first page of each paper) to the effect that the papers are in final form and will not be published elsewhere.  
Some but not all conference proceedings are refereed less thoroughly than reputable journal articles.  As a result, mathematicians are sometimes suspicious about results published in conference proceedings, and administrators sometimes assign less value to such publications.   Some conference proceedings have responded to this problem by explicitly saying (usually in the preface of the proceedings volume) that the papers have been refereed to the standards of such-and-such journal.  Nevertheless, I would advise young (= not yet tenured) mathematicians to publish most if not all of their work in regular (and reputable, of course) journals. Once you have tenure, so that administrators' opinions are less critical for your life, it becomes reasonable to contribute more to conference proceedings.
