Microsoft's Internet Explorer 8 (IE8) again trounced rival browsers in a test of their malware-blocking abilities, catching 81% of attack code-infected sites, according to a testing company.

IE8's skills at sniffing out malware sites improved by 17% since March, said Rick Moy, president of NSS Labs, the firm that conducted the benchmarks. The testing was sponsored by Microsoft's security team.

IE8's improvement, and its dominance over competitors, could make some users reconsider their decision to abandon Microsoft's browser for one of its challengers. "Should people rethink that decision?" Moy asked. "By [this] data, absolutely."

While IE8 blocked eight of 10 of the malware-distributing sites that NSS included in its 12-day test, the nearest competitor, Mozilla's Firefox 3.0, caught just 27% of the same sites. Apple's Safari 4.0 and Google's Chrome 2.0, meanwhile, blocked only 21% and 7% of the sites, respectively. Opera Software's browser properly identified only 1%.

"I think it comes down to resources and the focus of these companies," Moy said in an interview, referring to Microsoft's ability to out-spend rivals on such things as security research and malicious site investigations. "The more researchers you have, the better you'll do. Microsoft has a certain amount of paranoia [about security] because of its footprint of services that get attacked all the time, like Hotmail, and it has the money to hire really smart people."

Opera, which performed the poorest in the malware-blocking benchmarks, is an example on the other end of the spectrum, said Moy. "What resources do they really have to bring to the problem?" Moy said. "There's a lot that can't be solved with software, but requires the human element."

NSS tested five Windows-based browsers -- IE8, Firefox 3.0.11, Safari 4.0.2, Chrome 2.0.0.172.33 and Opera 10 beta -- against more than 2,100 malware sites in 69 test runs over 12 days. Like the tests NSS Labs ran last March, the sites were so-called "socially engineered" malware sites, the type that trick users into downloading attack code. Typically, the download is disguised, often as an update to popular software such as Adobe's Flash Player.

The tests did not include sites that launch "drive-by" attacks that don't require user interaction, an increasingly common tactic by hackers who often infect legitimate sites with kits that try a number of different exploits in the hope of compromising an unpatched browser or PC.

To defend against the kind of sites that NSS tested, browser makers have added anti-malware features to their software. Microsoft, for instance, has aggressively touted its SmartScreen Filter, a new malware-detection feature in IE8.

All browsers that include such a tool -- or anti-phishing tools, which operate in a similar fashion -- rely on a "blacklist" of some sort. The list, which includes known or suspected malware sites, is used to display warnings before a user reaches a site, but after the URL is typed in.

"The foundation is an in-the-cloud reputation-based system that scours the Internet for malicious sites," explained Moy, "then adds them to a black list or white list, or assigns them scores." The browser then uses that information to block or allow access to a site.

IE8 significantly improved its lead over other browsers since March, Moy noted, with its browser's malware-blocking rate up 12 percentage points -- a 17% improvement -- while rivals' scores declined across the board. Firefox dropped three percentage points, for example, as did Safari 4; Chrome fell eight percentage points and Opera, four.

Even though Firefox, Safari and Chrome all rely on the same data source for their anti-malware blacklists -- Google's SafeBrowsing API -- their scores varied considerably, something Moy thought was due to each browsers' use of the list. "Google produces the API, but that doesn't mean all the browsers consume the data in the same way at the same time," he said. "We don't have any visibility on how many people are looking at the [SafeBrowsing] data, but clearly Firefox must be adding other things to it."

Moy also said that IE8's anti-malware protection improved over time at a greater rate than did its rivals. Because NSS Labs tested every four hours, it was able to measure how quickly each browser reacted, and blocked, a new threat introduced into the test. While IE8's score jumped from 51% on Day Zero -- the day the infected site debuted on the Internet -- to 91% by Day 5 (a 40 point jump), Firefox was only able to muster a 10-point increase, from 14% to 24%. Chrome improved the most over the course, starting at just 3% on Day Zero and ending at 14% on Day 5.

"I was surprised when Microsoft got 69% in the first study," said Moy. "Then they went from 69% to 81." NSS hopes to repeat the test before the end of the year.

According to the most recent data from Web metrics vendor Net Applications, IE8 accounted for 12.5% of all browsers used in July, representing 18% of all versions of IE in use.

The NSS report can be downloaded from the company's Web site ( download PDF).