It's been discussed several times. Remember that when you compare two meters you are dealing with two sets of errors, so 30mg/dl could be one meter consistently reading 15mg/dl high and the other 15mg/dl low. In fact 15mg/dl is well within the accuracy limits of current standards and does, in fact, even meet the proposed new (2011) standards!
More important the accuracy of any meter decreases as the reading increases. In fact the manufacturers and the FDA typically quote a percentage accuracy, not an absolute one. At a blood sugar of 200mg/dl even a 30mg/dl error is just 15% wrong, and that is still within the current FDA guidelines for accuracy (+/- 20% - up to 40mg/dl wrong)!
The current standard is ISO15197:2003. While few people read ISO standards because of their enormous cost the current standard is documented here:
(You can also just Google for "ISO 15197" and look for the PDF file 120710-Hasslacher.pdf).
The accuracy of any meter is stated on the documentation that comes with it. As the above PDF documents meter accuracy is normally a lot better than the FDA standards, at least for higher readings. In fact when I've tested pairs of meters the consistent errors I've noticed have been below the variation in pairs of readings from the *same* meter. Particularly at high BG levels (above 200mg/dl) I see quite large differences (as much as 50mg/dl) between two readings done one after the other. I'm using a Contour USB, which comes out as pretty accurate in the Hasslacher tests and, indeed, 50mg/dl differences in 200mg/dl BG could be as little as 12.5% error (e.g. if the two readings were 175mg/dl and 225mg/dl).