BMI said:
As for the A123 fires I am almost certain these were "electrical" fires" rather than "chemical fires" (attributed to the chemical composition of the battery) so the A123 cells didn't burn because the LiFePo4 caught on fire but rather because of excessive discharge current (or short circuit) causing severe overheating of wires/insulation and surrounding flammable materials. A couple of the EV fires have started due to poor spot welds on the A123 cells which have fatigued over time resulting in less/smaller contact area where the battery tab is in contact with the cell which has caused excessive heating at the joints and ending up in a fire.
If you are going to be dragging more than 100amp peaks from such a small cell you want to make sure you have really good spot welds and preferably lots of them!
LiFePO4 can fail and burn like anything else. It was extensively proven during the Chev Volt torture testing with A123 cells, both 10Ah cylinder types (yes those exist) and 40Ah pouch type. The most safe chemistry tested was LiMnCo, which has a hard time doing much more than hiss and smoke a bit, no matter what sorts of cruel things you do like crush or stab metal rods through them or overcharge to 20v/cell. The A123 cells would burst into flames when you stabed into them, or when over charged to extreme levels.
You can read about some LiFePO4 battery fires that happened during charging.
Here is a LiFePO4 powered car:
http://www.batteryvehiclesociety.org.uk/forums/viewtopic.php?t=1825
http://www.rcgroups.com/forums/showthread.php?t=772399
Here is a little 15s2p A123 LiFePO4 pack in an RC helicopter that was set to charge at 4.5amps with a 3.6v cut-off.
There are many more references, but I don't want to side track this topic, so I will drop it

Any battery can burn, and not just during high current discharge or from electrical connection failure. Just takes the right conditions.