A couple things about the comparisons.
The tests of virus detection rate is really just a test of the deffinitions and hueristics. That means the test is essentially a snapshot (one moment) of what is in reality a long and on going race to catch up with virus creators as newer threats are created. The vast majority of actuall infections in the wild are limited to a small discrete number of viruses at any one time. It really makes no difference if a AV protects you from virus X if your chances of gettign that virus are in reality .000001% or some such... even if it is "in the wild". That vast majority of infections are as a result of the one or two big pandemics going arround at any one time. Those are the viruses that you need to be protected from right now. Therefore, what should be concerning you is the frequency with which the defs are put out. A number of the big name AV softwares update daily to keep on top of the big threats but have limited coverage of other in the wild, that are less prevelant and therfore less of a danger. Other AV Programs seem to have a higher rate of virus recognition overall. But this doesn't really help much in real life and will hurt you if they take a while between releases and def updates. If it takes more than 1-2 days for the updates then you are in danger I would say.
A more telling statistic, as opposed to the rate of detections on a per virus basis would be rate of detection based on the virus's numbers in the wild. Therefore if a virus is only infecting a very small number of computers relative to the total number of infections, that difference would be shown in the staistics. My guess is that if you drew up the nunbers that way you will see the numbers convergeing to somewhre arround 95%+ for all the antivirus software. That would give a far better snapshot view of how much protection you have in terms of stoping real infections that are active right now.
