Tim Rutherford-Johnson has a couple (here and here) of good items on the state of new music publishing with some lively comment threads.
Traditionally, music publishers had effective monopolies over music engraving and printing as well as distribution to local music shops. They had more-or-less efficient systems for dealing with rental catalogs and they had promotional capacity, both through schmoozing with prospective musicians and managers and through direct advertising for sheet music. Having made investments in their sheet music that could only amortize over long periods, they kept close control over their catalogs and inventory and they were often staffed with musically knowledgeable employees who could watch for errata and make sure that the materials sent out were the ones required to make a piece work. Moreover, getting published by a name house carried a professional caché, with which the road to tenure, for example, could be paved, and without which, one could be considered professionally deprecated.
None of this is true any more. Engraving and printing, although getting them done right remains an art, are no longer tightly-held secrets of the "industry". Every composer can, in principle, engrave and print her or his own work to a high level of quality. The local music shop with a large inventory of sheet music has basically gone the way of the dodo and was never that good, anyways, in ordering anything even slightly obscure. Publishers, always reluctant and sometimes even loathe to dealing directly with the end consumer of sheet music, have, for the most part, not warmed or gotten more efficient at the job and, again with a few exceptions, though they tend to be staffed with highly educated and truly music-loving musicians, they have downsized to the point where they only rarely can deal with errata or that second violin part with the impossible page turns. Advertising budgets do remain for a few houses, but it tends to be reinforcing the dwindling existing markets (i.e. to the handful of specialist journals and in festival programs) rather than reaching any significant mass (i.e. in the days when a Schirmer ad could be found in the Christian Science Monitor, right next to Nicolas Slonimsky's column on the kid's page.) Finally, the prestige of big name sheet music publication has diminished even in the most ivy-covered of tenure cases.
Instead, composers can pretty much go it alone (the great models here are Tom Johnson and Karlheinz Stockhausen) or in cooperation with colleagues (like Frog Peak or Material Press or Wandelweiser), with their own web sites serving as catalogs and ordering platforms for score delivery by post or email. The most immediate advantages for the composer are that he or she gets to keep all the license fees (instead of ceding the usual 50% to the publisher), has control over the quality of the materials produced, and has oversight over score sales and to some extent performances, recordings, and broadcasts. Scores in your own catalog will not get neglected or orphaned as they might in a publishing house that loses interest or get merged with or acquired by larger concerns. On the other hand, investment in score production and part extraction is all your own, you are on your own for promotion and you have to be available to answer inquiries and to send your materials out at any time (and yes, many score orders will come your way Sunday morning at 4:00 am, with a request that it be sent ASAP and preferably earlier than that.) The disadvantages also include not being able to use sales and license fees from the back catalog to subsidize new catalog items as well as having to organize promotional contacts and networks from scratch.
Clearly, each composer has to work out the balance of advantages and disadvantages between going with a traditional publisher and going it alone for herself or himself. Personally, the greater advantage for me is not to got to a traditional publisher, but if my catalog had more choral and school band or orchestra music, the distribution logistics for handling the volume might throw the balance in the other direction. (That said, it appears that a number of composers are now able to use direct sales of school ensemble music to earn a fair income, so if you can deal with a large volume, go for it.) In any case, if you do decide to go with a traditional publisher, it seems wise to insist on these terms as a minimum: (1) the composer should not have to pay in any cash up front for publication, especially when print-ready scores or parts are delivered by the composer, (b) the amount and form of promotion to be made by the publisher should be defined in detail in the contract and may be reflected in the publisher's share of license fees or sheet music sales/rentals, and (c) should promotion not be carried out by the terms of the contract or should the materials be orphaned by the publisher, all publication rights should revert to the composer.