Unit 3's MOX fuel contained plutonium, a very toxic as well as radioactive element (e.g., see here).
Nuclear engineer Arnie Gundersen of Fairewinds has hypothesized that unit 3 experienced a "prompt criticality" in the spent fuel pool.
Although Gundersen's account has been disputed, there has been no compelling evidence (in my opinion) to rule out his interpretation and, more importantly, reactor fuel from Fukushima's explosions was detected worldwide regardless of the explosion form (e.g., see http://www.datapoke.org/blog/89/study-modeling-fukushima-npp-p-239-and-np-239-atmospheric-dispersion/).
One group of researchers using Cesium-134 as a marker, which has a short half-life, reported detecting Fukushima plutonium, in Lithuania:
G. Lujanienė , S. Byčenkienė, P.P. Povinec, M. Gera M. (27 December 2011) ‘Radionuclides from the Fukushima Accident in the Air Over Lithuania: Measurement and Modeling Approaches’, Journal of Environmental Radioactivity, 114, 71-80.
My bet is that Diablo Canyon nuclear power plant in California, situated on a fault and next to the Pacific Ocean, will be the next Fukushima.
Fukushima today