Mozilla says on their Wiki that in order to improve Firefox's marketshare they'll need to improve their ability to retain users. It seems that only 50 percent of the people who download Firefox actually try it, and about 50% of those people become active users.
So basically 75% of the people who download Firefox don't become active users.
Firefox's claims to success have been supported by clocking up the number of downloads, encouraged by payments from Google, but it turns out that 75% of the people making those downloads don't "continue to use it actively."
Mozilla also has a 12-point plan to increase retention levels, though it's actually an 11-point plan with one item repeated.
Well, I'm not convinced that 25% is such a bad figure. I remove at least 90% of the programs I download and try, often within seconds. Sturgeon's Law applies.
But Mozilla does have a real problem, which is that the days of easy pickings are over. Microsoft's IE7 works well and so far hasn't had the sort of security problems that afflicted IE6. (In fact, Firefox suffered from more vulnerabilities than IE, in 2006, according to IBM.)
It also has competition from Apple's Steve Jobs, who has publicly committed to wiping Firefox out. Whether he's just blowing smoke remains to be seen.