Here’s a report (pdf) on the results of some simulations run by the RAND Corp. on the effects of a nuclear bomb that arrives in southern California in a shipping container. The short summary of the probable near- and long-term consequences: very bad.
Of course this is a simulation and it’s impossible to know how accurate it is. The attack it describes may never happen, and if it does happen the effects might not be as bad as predicted. Or we might be attacked in other ways. The more important questions, which the RAND simulation can’t address and which may not be accessible via simulations, are, what are the odds of a WMD attack in the first place? and if the odds are much greater than nil, how can we minimize them?