Dylan Matthews has a useful discussion of five of Coase's papers in the Washington Post (available here).
After Milton Friedman, Ronald Coase was probably the most important economist of the last 100 years. Given the biases of the Washington Post and particularly the wonkblog, I was very surprised that this discussion was as fair as it is.
Take the discussion in the first paper discussed (“The Problem of Social Cost”). "Coase theorizes that, in an ideal world with no transaction costs, the two could reach a bargain pleasing to both parties without going to court at all."
The Coase theorem really doesn't depend upon transaction cost being zero. No one, particularly Coase who wrote articles about the importance of transaction costs (see the second paper listed on "The Nature of the Firm"), believed transaction costs are zero. The point is that as long as the transaction costs are less than the gains from trade it doesn't matter who has the property rights, you will always get the efficient result. (Who gets the property rights does matter in terms of wealth and who gets the payments but not in terms of the final outcome of output.) When transaction costs are greater than the gains from trade, it does matter who has the property rights and they should be allocated in a way that would be consistent with who would pay the most for them.
The next three papers discussed seem at first glance to be pretty accurate.
A copy of Coase's Noble acceptance speech is available here.