I can't remember the details fully, but they tested baby car seats and found all but one to be severely lacking, but if I recall correctly they used insanely high speeds for the crash tests (at least double what cars are crash tested at) and I think the speeds might have also varied. Given that they admit blind, unfounded bias in favor of certain auto manufacturers, I wouldn't put it past them to also have a preferred car booster seat manufacturer that was blessed with some of the slower speed tests.
Edit: Straight from Consumer Reports
http://www.consumerreports.org/cro/b...w/0507seat.htm
So I was wrong, two seats made it through OK, not just one. Also I don't see anything in here about the varied speeds, so I don't know if that was mentioned in a different article on the incident or if I'm confusing that with something else entirely. The best part about all this is this happened almost at the same time that CR was forced to acknowledge they automatically grant Honda and Toyota the highest reliability marks regardless of what the data says.