Glucose monitors are quite accurate in that any given reading on the same sample of blood should only vary by a few percentage points (i.e., 5–10).
In order to obtain approval from the Food and Drug Administration, such medical devices have to show that they are both precise and accurate. This means that they will obtain a similar result under similar conditions and that the result will be close to the real true value, as best as we are able to ascertain it, i.e., according to the accepted gold standard, which is glucose measured in the laboratory.
Furthermore, when such a device is compared to the gold standard measurement in clinical trials, a specified number of readings have to fall very close to the readings obtained by the gold standard device. This value is more than 95%. The rules for approval of such devices have been published by the FDA and can be viewed at http://www.fda.gov/cdrh/oivd/guidance/1171.pdf.
Although users are often surprised by the variation in results that they will see upon obtaining a reading within a very short space of time from a prior reading, it has to be acknowledged that the variation is seldom such that a different course of action would be taken. For example, insulin would be given or not given or a low blood sugar level would be treated or not treated.
Actual numerical accuracy tends to be less when the glucose level is very low. However, since such a low value should always be treated, the accuracy of the actual reported number is not critical.
When a surprising glucose value that does not fit with the expected circumstances is obtained, it is always prudent to recheck that reading, on either the same or a different monitor.
Also, it is advisable to check that the monitor is functioning well, by checking the low battery indicator, assuring that the correct coding has been entered and that the test strips are within date and inserted properly.
Sometimes, variation in technique by the user accounts, at least partially, for variation in the result obtained. This can include liquid on the puncture site, contamination with glucose, excessive squeezing of the finger, and sweating.