Testing antivirus software is always a dicey process, but it's a worthwhile exercise. I've never seen a test as exhaustive as the one performed by this site, which put 147,000 virus samples on a PC and ran 58 virus scanner applications to see which was "best."

I say "best" because I don't think the best virus scanner is the one that necessarily mean the application that caught the most viruses. How can that be? Because 99.9% of viruses are not "in the wild" at any one time. Crooks know that most viruses are caught by experts in a few weeks or months and antivirus software is eventually updated to catch them. It just isn't worth the trouble to circulate an old virus.

The result is that only a few dozen viruses are typically circulating at any one time. This means that a good antivirus program doesn't necessarily have to catch every virus, it just has to catch all the ones circulating right now.

Now don't get me wrong: Catching all 147,000 viruses is good, but catching all 100 viruses that are current threats is even better. Since we don't know which viruses were caught by each scanner and which were not, the results on the virus.gr site should be taken with a grain of salt. To wit, here are the top five, along with the percent of viruses they caught: 

1. Kaspersky version 6.0.0.303 - 99.62%
2. Active Virus Shield by AOL version 6.0.0.299 - 99.62%
3. F-Secure 2006 version 6.12.90 - 96.86%
4. BitDefender Professional version 9 - 96.63%
5. CyberScrub version 1.0 - 95.98%

Some old guard names in there, but some surprises, too: An AOL virus scanner ties for #1? Wow.

Where are the big guns? McAfee's at #13 and Norton's at #22. Not great, but again, don't place too much stock in a test like this. What really concerns me is the stuff at the bottom of the list. I've never heard of Abacre 1.4, but with a 0.0% detection rating, I'm not rushing to install it.