I have worked in metrology in the aerospace industry for about 30 years and have calibrated thousands of torque wrenches.
Snap-On has very good equipment and the accuracy depends on which model you have. The statement above about accuracy is true, however, how do you know if its truly accurate if it is not tested on a consistant/regular basis?
If it has not been stored corectly (at its minimum setting) or if it has been dropped, or over torqued, will definately effect its accuracy.
Most click type torque wrenches are manufactured to an acuracy of +/-4% of indicated value in the C.W. direction, and +/-6% of indicated value in the C.C.W dir. Some of Snap-On's dial type wrenches have an accuracy of +/-2% ind.
Those mfg. accuracies are from 20% to 100% of their range, so in other words they are not intended to be used in the bottom 20% of their range.
Deflecting beam torque wrenches are OK, but like any wrench, you never really know unless you get it on a high quality, highly accurate load cell torque calibrating machine that is at least 4 time more accurate than what you are testing. Even then, they (deflecting beam wrenches) are not very repeatable.
I know its all relative, and most people dont get into this kind of detail, but it is my job, and the wrenches I calibrate are used on multi billion dollar launch vehicles and satelites.
Bottom line, get them calibrated if you really want to know that they are accurate.