As the very last impression your reader gets of your paper, the conclusion is your opportunity to sell your argument once and for all.
It's a place for reflection, for looking back at the relationship between the numerous ideas of your paper. Most importantly, however, it ought to be the site of your most complex analysis; that which incorporates everything that's gone before.
Some General Cautions
DON'T allow the conclusion to become merely a restatement of the thesis with a couple of linking sentences beforehand.
DON'T view it as merely an ornamental way to end your paper - its role should be to justify your paper at the highest level.
DO analyze how your argument has changed as your paper has progressed. If you haven't proven anything more than merely what you mentioned in your introduction, you haven't really said anything at all.
Throughout the course of a good paper new subtleties of argument ought to have manifested themselves, and the place to integrate all these subtleties into a new, more powerful statement of your thesis, is right in the conclusion.
DON'T begin your conclusion with the opener "In conclusion...". That makes your paper awkwardly self-conscious and contrived, rather than naturally unfolded.
DO attempt some sort of unified closure, with respect to what you set up in the introduction. If you used one of the previously mentioned clever introductions, make reference again to the quote, questions, or anecdote you incorporated.
DO consider linking your argument to a more universal idea, analyzing its relevance with an eye on the new angle your argument proved.
