I can only tell you what I do in my inorganic courses that tend to be small (under 20).  I don't have the problem sets count as much, say 5% of the total grade or less.  In class, I have students post their answers on the board by calling on them at random.  We then discuss their answers and any other answers there might be.  If necessary, we then discuss the right answer.  I encourage students to make notes on their problem sets but require them to use an different color pen/pencil.  I then collect them and review them on a very simple scale of check plus (everything right exceptionally good work), check (a few mistakes but general understanding), check minus (minimal effort) or zero (didn't hand it in).  Let them copy or cheat.  Well, ok, don't let them, but I don't worry about it too much.  It is only 5%.  If they aren't doing the work and think about the material, I am pretty confident I will kick their butts on the exam.

you could have specific guidelines in your syllabus (pass out an addendum since you've already started).  I do this for lab reports and other electronic materials I collect on Sakai.  After a few years getting 13 copies of "labreport1.doc" I put explicit file naming conventions in my lab manual, along the lines of "your lab report should be named according to the convention xyzlab1.doc, where xyz are your initials."  I then included a precautionary statment that said if I received a file named incorrectly, it would not be graded.  Problem solved. 

For homeworks, you could say, 8.5x11 single sided, one problem per page, stapled in the top left, name on each page. 

Adam

This isn't really useful for *homework* sets, where you really want them to be able to work on a dead tree, but we've been immensely happy with giving students a template (in Word) for lab reports. The students like it because of the clear expectations (here's the formatting, here's the font size and spacing, here's the expected length), and we like it because it gives us a reproducible format to grade. They also like that this mimics the format of a real journal article. Sometimes they're shocked that we chemists do a lot of our own typesetting...I tell them that there's nobody more willing to work for free than a chem prof who is trying to get a paper accepted or a grant funded. For problem sets, I like the "tell them exactly what you want" approach. When I'm trying to get students to follow a format, I try to put it at the top of the assignment page. Due XX/XX/XXXX, All assignments must have a name in the top right corner, a list of students with whom you collaborate (I call them "your co-workers" in class) should be right underneath, the assignment should be turned in at the end of the period, and must be on 8.5x11" sheets of dead tree. If the format is up front and clear, students *like* it. If they have to go looking for it in the syllabus, they *hate* it, because it feels like "gotcha".

Back when I was a student grader...if a homework set wasn't stapled, we would find a large object like a block of scrapwood and staple their assignment to it about 72 times, such that they'd have to remove all the staples to see what they got on the set. I don't recommend it, but it does work...

I have done a similar thing (again with smaller classes) where I discuss problems/answers in class, let students add to answers in different color pen, then grade with a check plus/check/check minus system.  I agree that they will get out of the assignments what they put into them and that it will show on the exam.  This semester I'm teaching gen chem - so haven't quite decided how I will grade them.  I'm working on figuring out a way to use "clicker" questions based on the homeworks to assess how they're doing and give them credit for just participating rather than right answers.