How to Tell If Gold Is Real: 8 Tests Ranked
What each home test actually proves, which ones are worthless, and the two that will not lie to you.
Aktualisiert July 29, 2026
Test 1: read the hallmark (free, first step, not proof)
Start with the stamp. Solid gold is almost always marked, usually with millesimal fineness: 999 for 24K, 916 for 22K, 875 for 21K, 750 for 18K, 585 for 14K, 417 for 10K, 375 for 9K. Look inside ring shanks, on chain clasps or the small tag next to them, and on the back of pendants and earring posts. A loupe or your phone camera at maximum zoom will read marks too small for the naked eye. Marks that mean not solid gold: GP and GEP (gold plated), HGE (heavy gold electroplate), RGP (rolled gold plate), GF and fractional forms like 1/20 12K GF (gold filled), and vermeil (gold over sterling silver). Also treat "18KGP" as plated - the GP governs. The limits of this test are important. Stamps are trivially easy to forge, and counterfeit jewelry is routinely stamped correctly. An absent stamp does not prove a fake either - marks wear off, get polished away, or were never applied on older and handmade pieces. A hallmark tells you what the piece claims to be, nothing more.
Test 2: the magnet test (free, rules out, never confirms)
Gold is not magnetic. Neither is copper, silver, aluminium, brass, lead, or the zinc in most gold alloys. Steel and nickel are. So a strong neodymium magnet - not a fridge magnet - will reveal a steel core or a nickel-heavy base metal. If the piece jumps to the magnet, it is not solid gold. That is a genuine and useful result. Watch for weak attraction too, which can indicate a steel core under a gold shell or a steel spring inside a clasp. But passing the magnet test proves nothing at all. Brass, bronze, copper and lead-filled fakes are all non-magnetic, and gold-plated brass is the most common counterfeit material there is. Every serious fake passes this test. Use it as a five-second filter, then move on.
Test 3: visual and wear inspection (free, surprisingly effective)
Plating wears through at friction points, and this is where most fakes give themselves away. Check the high-wear areas: the underside of ring shanks, chain links where they rub, clasp edges, earring posts, and any raised detail. A different colour showing through - silvery white, pink, or green-grey - means plating over base metal. Look for green or black discolouration on skin contact areas, which comes from copper or brass oxidising. Look for a colour that is too orange, too brassy, or unnaturally uniform. Real gold has a soft, slightly warm yellow that does not change tone across the piece. Check construction quality: sloppy solder joints, seams, pitting, casting bubbles and mismatched components are common on fakes and rare on genuine work. A magnet-safe piece with worn-through plating at the clasp is settled - it is plated.
Test 4: density measurement (free, and the best home test there is)
This is the one home test that can actually distinguish gold from most fakes, and it needs nothing but a 0.01 gram scale, a cup of water and some thread. Gold is extraordinarily dense - 19.3 g/cm³ pure - and almost nothing that looks like gold comes close. Method: weigh the piece dry and record it. Fill a cup with water, place it on the scale, and tare to zero. Suspend the piece by thread so it is fully submerged but touching neither the sides nor the bottom. The new reading in grams equals the volume of the piece in cubic centimetres, because water is 1 g/cm³. Divide dry weight by that volume to get density. What to expect: 24K is 19.3, 22K around 17.7-17.8, 18K yellow around 15.2-15.9, 14K yellow around 13.0-13.6, 10K around 11.5-11.9. Brass is 8.4-8.7, copper 8.96, steel 7.8-8.0, and silver 10.5. A reading of 8.5 on something stamped 585 is a settled answer. The caveats: it fails on hollow pieces, on anything with stones or non-gold components, and on very light items where scale error dominates - stay above about 3 grams. It also cannot distinguish gold from tungsten, which is 19.25 g/cm³ and is the reason tungsten-cored bars exist. For jewelry, though, this test resolves the overwhelming majority of cases.
Test 5: acid testing (cheap, destructive, reasonably accurate)
An acid test kit costs about as much as a takeaway meal and contains graded nitric and hydrochloric acid solutions for 10K, 14K, 18K and 22K. The principle is that gold resists acid and base metals do not, and stronger solutions are needed to dissolve higher karats. Method: file or scratch a small inconspicuous spot to expose metal below any plating - this is essential, because testing the surface of a plated piece just tests the plating. Apply a drop of the karat solution you expect. If the metal dissolves, fizzes or discolours, it is below that karat. If it stays unchanged, it is at or above it. Work up from the lowest solution to find the boundary. The scratch-plate variant is gentler: rub the piece on a testing stone to leave a metal streak, then apply acid to the streak and compare against streaks from known reference needles. Downsides: it marks the piece, the acids are corrosive and need ventilation and eye protection, results are a band rather than a number, and it can be fooled by thick plating if you do not file deep enough. Do not acid test anything you might want to sell as jewelry rather than scrap.
Test 6: XRF analysis (definitive, non-destructive, usually free)
X-ray fluorescence spectrometry is what the trade actually uses. The machine bombards the surface with X-rays and reads the characteristic emission of each element present, returning a full alloy breakdown - gold, silver, copper, zinc, nickel, palladium - in a few seconds, with no damage to the piece. It is accurate to within a fraction of a percent on solid material, and it is the only test that gives you a number rather than a band. Most jewelers, coin dealers and refiners have one, and many will test a piece for free or a nominal fee. This is the answer when the stakes justify a trip. Its one real limitation is depth: XRF reads a surface layer measured in microns, so heavy plating can defeat it if the operator does not also file a test window or cross-check with density. Reputable buyers do both, which is why testing done in front of you with a visible reading is a good sign about who you are dealing with.
Tests that do not work
The ceramic scratch test. Dragging the piece across unglazed ceramic supposedly leaves a gold streak if real and a black streak if fake. In practice plenty of base metals leave gold-coloured streaks and plating leaves one too. It also scratches your item. Skip it. The bite test. Gold is soft, so teeth supposedly leave marks. Lead is softer, tin is softer, and 10K gold is hard enough to resist. You will damage a real piece and learn nothing about a fake one. This is folklore, not a test. The float test. Gold sinks. So does brass, steel, copper and almost every other metal. This only rules out hollow plastic. Vinegar, bleach and makeup tests. Household chemistry produces inconsistent results, can permanently discolour genuine alloys, and cannot distinguish 10K from 18K. Bleach on gold-filled will attack the base metal at the seams and ruin the piece. The "it feels heavy" test. Human weight perception is unreliable and heavily influenced by size and expectation. Use a scale and do the density calculation instead - it takes two minutes and gives you a number.
A sensible testing order
Run them in increasing cost: hallmark inspection, then magnet, then wear inspection, then density. That sequence is free, takes under ten minutes, and resolves most questions on solid jewelry over three grams. If density is ambiguous, or the piece is hollow, stone-set or very light, escalate. Acid test if you are prepared to mark it and it is destined for scrap anyway. Take it for XRF if it might have value intact, if the amount at stake is meaningful, or if you simply want a number you can rely on. One thing worth remembering throughout: authenticating gold and valuing gold are different questions. A piece can be genuine 14K and still be worth little because it is light or hollow, and a piece can be genuine and worth far more than melt because of who made it.
Wichtige Punkte
No free home test can distinguish solid gold from a well-made plated fake with certainty. The magnet test only rules things out - passing it proves nothing. Density measurement is the best home test and needs only a scale and water. XRF analysis is definitive, non-destructive, and most gold buyers will do it free.