According to von Hippel:
Separated plutonium, being only weakly radioactive, is easily carried off—whereas the plutonium in spent fuel is mixed with fission products that emit lethal gamma rays. Because of its great radioactivity, spent fuel can be transported only inside casks weighing tens of tons, and its plutonium can only be recovered with great difficulty, typically behind thick shielding using sophisticated, remotely operated equipment. So unseparated plutonium in spent fuel poses a far smaller risk of ending up in the wrong hands.
While this is certainly true (how would you steal a whole spent fuel assembly?), is separated reactor-grade plutonium really all that benign?
According to the World Nuclear Association,
The approximately 1.15% of plutonium in the spent fuel removed from a commercial LWR power reactor (burn-up of 42 GWd/t) consists of about 53% Pu 239, 25% Pu-240, 15% Pu-241, 5% Pu-242 and 2% of Pu-238 which is the main source of heat & radioactivity. Comparable isotopic ratios are found in the spent fuel of CANDU heavy-water reactors at much lower burnups (8 GWd/t), due to their use of natural uranium fuel and high thermal neutron spectrum. (From gas graphite Magnox reactors the plutonium has more Pu-239 - about 65%, plus 25% Pu-240, 5% Pu-241, 1% Pu-242 and negligible Pu-238.) Reactor-grade plutonium is defined as that with 19% or more of Pu-240.
That 15% Pu-241 turns out to be a big deal, as this isotope has a half-life of only 14.3 years, and it decays into Americanium-241--which is a fairly strong gamma emitter. As the WNA notes:
It is of interest (and some significance in recycling spent fuel) that if too much Am-241 builds up in plutonium separ-ated from spent fuel, it cannot readily be used for mixed oxide (MOX) fuel because it is too radioactive for handling in the normal MOX plant. For instance, British Nuclear Fuels at Sellafield, UK, can handle plutonium with up to 3% Am-241, hence up to 6 years old (any more would need special shielding).
Obviously, the radiation hazard posed by Am-241 is non-trivial. Just how radioactive is the amount of material needed to build a "nominal" bomb?
Von Hippel implies in the article that the amount of civilian plutonium needed for a bomb is 8kg. If the plutonium originally had the isotopic makeup given by the WNA for LWR fuel, after 7.15 years it would be 7.5% Am-241--about 600 grams of Americanium. For comparison, one gram of Am-241 is more than sufficient to build 5000 smoke detectors--so that 8kg of material produces gamma emissions equivalent to more than three million smoke detectors! (It's around 105 GBq, or a little under 3 Ci, if you'd like a figure.) The total radioactivity of the 8kg is far greater than this, but it's far easier to shield against beta and alpha radiation than gamma. The whole package would also produce a lot of heat--well in excess of 100 watts. In short, this is not something you would want to hide in your pants.
The upside of this is that Am-241 contamination makes it possible to design very robust security measures against theft of the material. Anti-proliferation types tend to worry a great deal about the fact that meaningful amounts of plutonium end up getting lost in reprocessing equipment, so precise accounting of all seperated plutonium produced is impossible. But I don't really understand why this is a problem, as because of Am-241 contamination, sensible radiological control measures at all possible exits should make it impossible to smuggle even very limited amounts of plutonium from reprocessing and storage facilities. Given that sensitive radiological monitoring equipment is very proven technology, this should be quite easy to achieve. Even in very "fresh" separated plutonium, the gamma radiation should be readily detectable. Just make sure the thief can't run out a back door or jump out a window or something, and preventing material theft is straightforward.
Of course, there is also the question of whether such "reactor-grade plutonium" can really be built into bombs. Much ink has been spilled over this issue, including by Amory Lovins and von Hippel himself. I feel that most of the writing on this issue obscures the real issues and performs its analyses in complete isolation from the realities of nuclear weapons design. In particular, they tend to ignore the fact that even though it's quite possible to build nuclear weapons out of non-weapons grade plutonium, the stuff produced in LWRs is probably unsalvageable. While it could be used in bombs by sophisticated states with a fairly advanced knowledge of nuclear weapons design, it would be very difficult for rogue states or terrorists to fashion it into working devices.
Weapons-grade plutonium contains at least 90% Pu-239; American nuclear weapons apparently use approximately 94% Pu-239. It's true that in 1962 the United States tested a nuclear weapon that used plutonium from a British Magnox power reactor, but according to the WNA, "The isotopic composition of this plutonium has not been officially disclosed, but it was evidently about 90% Pu-239." According to Carey Sublette, however, this weapon used >19% Pu-240. I'm not sure who to believe, but it is known that this weapon achieved only nominal performance using what was probably a very sophisticated design. In any case, this is the only known weapon ever tested that was built using "reactor-grade" material. My suspicion is that this is for good reason. Nuclear weapons states have devoted enormous resources to nuclear weapons research; my guess would be that the use of reactor-grade plutonium in weapons has been studied thoroughly in many countries, and that it has been universally rejected.
Where this become interesting is in the case of aspiring nuclear powers to whom the idea of a clandestine nuclear weapons program using reactor-grade material would be particularly attractive. The obvious example here is Taiwan, which developed a substantial civilian nuclear infrastructure while simultaneously seriously mulling the acquisition of nuclear weapons. However, prospective nuclear states do not seem to be attracted to the idea of building bombs out of this material. But I bet they've thought about it, and that some very knowledgeable people have devoted many hours to analyzing the idea. In fact, the reason that I think that the proliferation risks of weapons-grade plutonium are overblown is because apparently these people all came to the conclusion that they'd be better off developing infrastructure that can build better weapons. There are many reasons why--one of which being that nuclear weapons are generally acquired for their deterrent value, and apparently unreliable low-yield weapons of the sort that can be built from reactor-grade plutonium just won't do for this.
But in any case, I don't see why security of separated plutonium stocks is an unsolvable problem. Although I'm in favor of proliferation-resistant thorium fuel cycles, I don't think that proliferation risks are necessarily such a strong argument against reprocessing. The problem with reprocessing is that uranium is cheap, as is dry cask storage for spent fuel. Until those factors change, it seems silly to invest in very expensive reprocessing facilities. We can always reprocess the spent fuel into fuel salt for MSRs later.