Jeffrey Goldberg's article in the latest Atlantic, on whether Israel will (or should) attack Iran's nuclear facilities in the coming months, is the best article I've read on the subject—shrewd and balanced reporting combined with sophisticated analysis of the tangled strategic dilemmas.
Whatever you think should be done about the Iranian program to build an A-bomb (and Goldberg describes his own position as one of "deep, paralyzing ambivalence"), read his piece before thinking about it much more.
Based on interviews with dozens of Israeli, Arab, and U.S. officials, Goldberg puts the odds of an Israeli strike by next July—involving 100 or so F-15E, F-16I, and F-16C aircraft dropping munitions on the uranium-enrichment facilities at Natanz and Qom, the nuclear-research center at Esfahan, and maybe the Bushehr reactor, among other sites—at "better than 50 percent."
He fully itemizes the risks and possible catastrophes of such a move: lethal reprisals from Hezbollah, if not Iran itself; a full-blown regional war; a cataclysmic spike in oil prices; a rupture of U.S.-Israeli relations; a rash of terrorist strikes against Jews worldwide; and—not least and most likely—a solidification of the mullahs' rule in Tehran.
Yet there are also considerable risks in letting Iran go ahead and build a small nuclear arsenal. I don't think (though many Israelis, understandably, do) that the mullahs would nuke Jerusalem, once they had the means to do so; Israel has about 100 A-bombs and the means to deliver an obliterating response. (The mullahs may finance suicide bombers, but they aren't so suicidal themselves.)
Still, a nuclear-armed Iran would provide cover (a "nuclear umbrella" of sorts) for Hezbollah and other militant proxies to step up their aggressiveness; it may draw smaller countries in the region into Iran's orbit (and certainly deter them from doing anything against Iranian interests); it could sap the credibility of a subsequent U.S. policy to "contain" Iran (if we declined to use force to stop Iran from building a bomb, some might doubt we'd use force to stop it from using one); and, for this reason, it could spur others in the region to build their own bombs, thus sparking a new nuclear arms race.
Some nuclear-deterrence theorists believe that an arms race might not be a bad thing. If several powers in a region all have nukes, the argument goes, they will be deterred not only from starting a nuclear war but also from starting a conventional war, out of fear that it could easily escalate to a nuclear one. This is one big reason there's been no war between India and Pakistan, or Russia and China, or (during the Cold War years) NATO and the Warsaw Pact.
There is something to this argument, but there are also at least three fallacies. First, all those pairs of rivals have come frighteningly close to nuclear war at various times and, in some instances, were saved as much by luck as by the doctrine of mutually assured destruction.
Second, all those countries' arsenals have been fitted with "permissive action links" and other security measures (in some cases with U.S. assistance) that minimize the chances of inside lunatics launching missiles without authorization. This may not be true in the case of a Middle Eastern arms race.
Third, Iran and Israel are so near each other as to make "warning time" of an attack almost impossible. Therefore, in a crisis, one side might launch a first strike, if just to pre-empt the other side from launching a first strike. (This is what deterrence theorists call "crisis-instability.") Even if neither side really wanted to nuke the other, circumstances might leave them with seemingly no alternative.