Like about two-thirds low.
The problem are the "red dwarf" stars, which are smaller and dimmer than our own sun. When such stars are relatively close by, they are not too hard to spot, either by direct observation or by their gravitational effects on companion stars. "Relatively close by" is the key phrase here -- it means that we can see red dwarf stars in our own galaxy, but we have a hard time seeing them in other galaxies. In our own Milky Way, red dwarf stars outnumber stars like our sun by about 100 to one. Unable to see or measure their presence in other galaxies, scientists pretty much had to assume that these cosmic low-watt bulbs were about as common everywhere else as they are here.
But some measurements suggested that other kinds of galaxies than our own might have more red dwarf stars in them. So astronomers tailored a telescope in Hawaii (scientists seem to be pretty smart about where they work) to look for evidence of them in distant galaxies. If there were in fact many more such stars in other galaxies as there are here, then their greater numbers could combine for a greater output, even though each individual red dwarf star would still be very dim. Nevada senator Harry Reid and Texas congressman Ron Paul used a similar principle to win re-election last month.
And behold, there were in fact not a hundred red dwarf stars for every sun-like star in some of those galaxies, but closer to a thousand instead. When astronomers sorted out how many of those kinds of galaxies they thought there were and how many new red-dwarf stars they might have, they figured that the universe might have three times as many stars as they would have guessed just a year ago.
Of course, the scientists involved caution against making too many predictions based on this new data. After all, the underestimate came from figuring that all the other galaxies had the same number of red dwarf stars our galaxy has. Nothing says that the galaxies they looked at are any more typical than ours was, so they say that dreaming up new theories or revising old ones based on this preliminary information would be a little premature. So they've learned that much, at least.