If Iran has a credible means to attacking Israeli cities with nuclear weapons, then Israel is constrained in their options should there be a dispute between the nations. For example, Iranian backed low level terrorism would likely continue against Israel under those circumstances, but if something particularily bad happens, you're a lot less likely to see an Israeli airstrike against Iran. Today, the Israeli Air Force could hit Iran pretty much anytime it wants will little chance that Iran can do anything about it. Likewise, should another big ground war like the 1973 Yom Kippur War break out, Israel might have a harder time of it if the other side has nuclear weapons. It might constrain how they mass their ground forces. That was one of the bigger problems with NATO contingency planning during the Cold War days. It was widely feared that the situation might quickly go nuclear against massed forces.

There has been multiple fire tests of the Arrow but I have no idea how many simultaneous launches it would take to saturate Arrow. That's probably one of the most closely held secrets of the system. Defending against the relatively slow and short ranged Scuds is one thing. Defending against the longer ranged and faster Shahab-3 that Tom linked to would be harder. Again, the performace capabilities of the Arrow are closely held secrets for good reasons, but it was specifically designed to be able to protect larger areas on the ground and to hit faster missiles than the Patriot, at least the PAC-II. I don't know how Arrow's capabilities compare to the newer Patriot PAC-III.

As for decoys, they cease to be a problem if you can wait until the RVs begin entering the atmosphere. Lightweight decoys will decelerate very rapidly when entering the atmosphere compared to actual warheads, so discrimination is much easier. If the decoys are as heavy as the warheads, then you might as well make them warheads and have a MIRV capability.

The US antimissile system is considering a different approach to target discrimination - launch multiple kill vehicles and let each one take out the most likely targets. I've read about really small homing vehicles that would only weigh a few kilos each, so the booster missile can carry a lot of them. This may well prove a more effective approach than using exotic radar or particle based discrimination systems. If memory serves, they plan to test this system in 2006.