A great bargain for a cheap scale. It will perform reasonably well if used carefully. Do not expect it to compete in accuracy or durability against laboratory scales, such as those by Mettler or Sartorius selling for $5-10K.This is a nice little scale with a 4 by 4 inch footprint, about 1-1/2 inch tall, including the clear and colorless hinged draft shield. It fits into a neat magnetic-latch box. Batteries are included, two 10g reference weights, and cheap tweezers. The weighing pan is non-magnetic stainless steel. I have two of these scales to save trips to the weighing room. The first one has been in use for 6 months with no problems.The scale is not quite ready to use as delivered. First, the weighing pan has adhesive residue. It is that sticky gum that can only be removed effectively with citrus oil or a citrus oil cleaner, such as Goo Gone. Second, and as others have noted, the pan sits too low on the scale body, and often touches the body of the scale during weighing. This, of course, will not work. The remedy is to glue a 10 mm diameter by about 0.5 mm thick non-magnetic metal washer to the underside of the pan. See my photograph at top of the Amazon listing. Although this will reduce the weighing capacity of the scale by the weight of the washer, it is insignificant. The washer I used weighed about 530 mg. Third, the tweezers are crude. If you have any serious weighing to do, you really should buy some quality tweezers with about tip #2. Dumont (Swiss) and others make good stuff. Of course, a single quality tweezer can easily cost more than this scale. You can economize by buying their epoxy coated electronic tweezers.NOW, REAL TESTING. How accurate? How precise?To test, I used a set of laboratory reference weights, and an older Mettler Toledo AG204 DeltaRange scale in the lab. This is an expensive machine (about $5-10K depending on features) that measures to 0.1 mg.First off, on the Mettler, the two 10 g AWS calibration weights weighed in at 9.9991 g and 10.0001 g. Not bad; they were within a milligram.How about precision? Took a 1000 mg lab reference weight and weighed it on the AWS ten times, with a complete on/off/tare between each weighing and a closed draft hood, in mg: 999, 1003,1001, 1001, 1000, 1000, 998, 998, 1000, 998. Not too bad. Variation of 5 mg. Perhaps with careful placement, a solid surface, and a quiet environment, this variation could be reduced.How about accuracy? Here are side-by-side measurements of reference weights on the Mettler and the AWS. The scales were not tared each time. Just put the weight on the pan. The AWS was not calibrated first trial, calibrated on the second trial. Accuracy could surely be improved with taring, stable environment, etc.Calibration weights, in mg - Mettler - uncalibrated AWS (calibrated AWS)5 - 5.0 - 5 (5)10 - 10.0 - 10 (10)20 - 20.0 - 19 (19)50 - 50.0 - 48 (50)100 - 100.0 - 100 (100)200 - 200.1 - 200 (199)500 - 499.8 - 501 (499)1000 - 1000.3 - 1001 (999)2000 - 2000.0 - 2001 (2000)5000 - 5000.2 - 5005 (5000)10000 - 9999.6 - 10011 (9999)20000 - 20000.0 - 20022 (20002)Again, this is impressive. With calibration, the measurements were off by no more than a milligram or two.My recommendation is that if you need an inexpensive scale to measure within a few milligrams, and are willing to calibrate manually, buy this scale. If you must have absolute accuracy, set aside at least $10-20K on a laboratory scale, stand, and accessories, and dedicated weighing room. And if you need real accuracy, get an equal-arm balance.Update 30 May 2012These two scales have been in use for over a year. Still perform well. They have been repeatedly tested. The measurements continue to be at least as accurate and precise as given above. Measurements have been published in peer-reviewed scientific journals. Depending on the project needs, and with reference weights and quality scales to verify performance, you can use this scale for scientific research.And remember, when you are weighing to the milligram and below, even the best scale will perform poorly with bad technique. This is not some yard-dog scale at a metal recycling plant. Have a clean, stable environment. Eliminate vibration and movement. Place the scale on a stable bench (if possible, granite is used in labs). You should be sitting, not standing. Arms off the bench. Eliminate air currents. No magnets. Keep magnetic steels away. Do not breathe directly on the scale. Use the draft shield. Gently place the object to be weighed on the pan. Don't drop it onto the pan. Don't use your fingers. Do not place the object directly on the pan. Use weighing paper or a disposable pan. After placing it on the pan and closing the draft shield, wait 10 seconds or so, for the scale to settle, before recording the weight. If you would like a fuller explanation of good weighing technique, Mettler has an excellent pdf user guide, "Weighing the right way," on their website.Update 4 Sept 2012Some mention should be made of the pan. It is not the usual flat pan, but rather a dish with a spout. It gives the unfortunate impression that one can load a powder or some such in the dish and then pour it out. Not standard practice. For my use the pan is fine, but others should consider a substitute "pan" out of non-magnetic metal or even cardboard with a washer spacer below. The subject powder or object should be placed on weighing paper or weighing cup. Be sure that there is enough room to close the draft hood, because without the draft hood closed, all milligram and sub-milligram scales are unreliable.Update 9 Sept 2012For those who have not worked in a lab, some mention should be made, after reading reviews on this scale, just what can be expected from any milligram or submilligram scale, whether it costs $25 or several thousand dollars. At this measurement level, subtle influences can greatly affect measurement. You cannot just throw a sample on the pan and hope for the best. Unfortunately, you need to know what follows.First off, these are scales, not balances. Balances compare a known, measured, mass against an unknown mass. They operate correctly anywhere there is gravity, no matter what the variation, even on the moon. In contrast, the AWS scale and even many expensive laboratory scales measure the force that gravity and centrifugal force of the earth's rotation exerts on a mass. Because gravitational force and centrifugal force vary slightly with the position of the scale on the earth, all these scales need to be calibrated after installation. Expensive scales do this automatically. With cheap scales, such as the AWS, you do this with a calibration weight. Unfortunately, the AWS calibration weight is too big, so you cannot close the draft shield when calibrating. Bad design. Use an inverted small beaker or such as a draft shield.Beyond that, there are several other significant sources of error affecting all scales:1. Drafts. All drafts cause severe variation in recorded weight. Keep away from windows, doors, HVAC outlets, radiators, etc. Use the draft shield.2. Vibration and movement. This will cause erratic readings and visible drift. Even absent outside movements, expensive scales take several seconds to settle down.3. Temperature. Expensive scales measure and compensate for temperature, Cheap scales do not. If you have access to a set of milligram reference weights, you can check for this.4. Humidity. Not too high, not too low.5. Light. Light causes heat, which changes the measurement. Keep out of sunlight and bright artificial lights.6. Leveling. Expensive scales have levels. Cheap scales do not.7. Sample position on pan. Always place sample in the middle of the pan.8. Electrostatic charge. A common and severe problem with low humidity and non-conductive weighing vessels, such as glass and plastic. Increase the humidity or change the weighting vessel.9. Temperature gradients. Differences in pan, weighing vessel, and sample temperature causes severe problems because of convection air currents.10. Magnetism. Obviously, any magnetic attraction between sample and something else will cause problems. Several materials commonly thought of as non-magnetic are somewhat magnetic.11. Buoyancy. Because of the sample size, the AWS scale should not be affected much by this, unless one is weighing a large, low-density sample, which would be buoyed by air.12. Oil and dirt. A common problem with beginners. Don't touch the pan or reference weights with fingers; use a tool, such as forceps or tweezers. Use weighing paper or some sort of weighing vessel on top of the pan. Generally, do not place a sample directly on the pan. If you make a mess, Windex is the usual cleaner for scales.So, for those reviews that claim without qualification that the AWS is accurate down to a milligram, it would take a lot of convincing. On the other hand, for those reviews that claim complete unreliability, assuming the item is not defective, it may just be one of the many inherent weighing problems.Still, a remarkable scale for the price.