How to Tell Real Gold From Fake: 7 Reliable Tests
By Sukie Gao · September 3, 2026Learning how to tell real gold from fake is a binary problem, and binary problems have a merciful property: you don't need a laboratory, you need a sequence. Somewhere in a drawer there is a chain, a class ring, a pendant from a relative, or a bracelet bought on vacation from a man who was very persuasive. It is either gold or it is a gold-colored object, and the difference is roughly the difference between $900 and $9. Most people resolve that question by squinting at it, feeling its weight, and deciding based on vibes. Vibes are exactly what counterfeiters design around — brass polishes to the same warm yellow, tungsten weighs the same in the palm, and a plated chain sounds identical when you drop it on a table.
What actually works is stacking cheap tests so each one covers the previous one's blind spot. Seven tests, run in order, will correctly sort almost everything a household will ever encounter, and five of the seven are free. The order matters more than any individual test, because each step is designed to eliminate a whole category of fake before you spend money or leave a scratch. Along the way you'll learn to recognize the three genuinely different things people call "fake gold" — plated, filled, and outright counterfeit — because the tests that catch one often miss another. By the end you'll have an answer you'd be willing to bet on, and if it's yes, you can put a number on it with a gold calculator before anyone makes you an offer.
Real vs. Fake Is Not the Same Question as "What Karat?"
People searching for how to tell real gold from fake are usually asking one of two very different questions, and running the wrong investigation wastes an afternoon.
This page handles the binary question: is this metal gold at all, or is it a base-metal object wearing a gold jacket? That's what you need when a piece is unmarked, when the stamp looks suspicious, when the item came from a flea market or a street vendor, or when you inherited something and nobody in the family knows its history.
The purity question — this is definitely gold, but is it 10K or 18K? — is a separate exercise, and our guide to how to test gold at home is the one built for it. That page ranks measurement methods by cost, from free hallmark reading to a $25 acid kit, and its goal is to land you on a karat number you can plug into a valuation. Use it *after* this page, not instead of it.
The practical rule: if you're not sure the metal is gold, start here. If you're sure it's gold but not sure how much, go there. Running purity tests on a plated piece produces a confident, precise, completely wrong answer — the acid reads the skin, the skin is real gold, and the brass underneath keeps its secret.
Test 1 — Audit the Stamp, Then Assume It Is Lying
Find the mark before you touch anything. On rings it hides inside the shank; on chains it's on the tag beside the clasp or on the clasp itself; on pendants it's on the bail or the back edge; on earrings it's on the post. A phone camera zoomed all the way in beats most reading glasses.
What you find falls into three buckets. Karat and fineness marks — 10K, 14K, 18K, 22K, or the millesimal numbers 375, 417, 585, 750, 917, 999 — are a claim that the piece is solid gold of that purity. Plating and filled codes are a confession: GP, GEP, and HGE mean electroplated; GF and RGP mean gold-filled or rolled gold plate; a fraction in front, like 1/20 12K GF, tells you the gold layer is one-twentieth of the item's weight. Foreign or unfamiliar marks need the decoder in our guide to gold hallmarks before you conclude anything.
Then distrust all of it. Stamping dies cost less than a dinner out, and the most common counterfeit jewelry in circulation is brass wearing a crisp, confident 14K. A stamp raises or lowers your prior; it never closes the case. Two tells worth noting: on genuine pieces the mark is usually crisp and squarely oriented, while faked stamps often sit crooked, shallow, or oddly deep, as if pressed by someone in a hurry. And an unmarked piece is not automatically fake — plenty of pre-1970s American jewelry and most handmade work was never stamped at all.
Test 2 — The Magnet, Which Only Half Works
Gold has no magnetic response whatsoever. Run a strong magnet over a piece and a firm tug means there is steel or iron inside it, which means the piece is not solid gold — full stop, no appeals.
Use a neodymium magnet from a hardware or craft store, not the one holding a takeout menu to your refrigerator. Lay the item on a table or dangle it from a thread, bring the magnet within a finger's width, and watch for movement. Test the body of the piece specifically: clasp springs, watch pins, and earring backs frequently contain steel even on unquestionably genuine gold, so a slight pull at the hardware alone means nothing.
The limitation is severe and worth internalizing. Brass, bronze, copper, zinc, lead, aluminum, silver, and tungsten are all non-magnetic. Every serious fake ignores a magnet. What the test buys you is speed: in four seconds it clears out the enormous population of steel-cored costume chains and plated fashion jewelry, and that's a real category. Read the result asymmetrically — sticking is proof of fake, not sticking is permission to continue.
Test 3 — Density, the Number That Is Genuinely Hard to Fake
This is the most powerful free test, and the one almost nobody bothers to run because it requires arithmetic.
Pure gold has a density of 19.32 g/cm³. Alloys come down from there — 18K lands roughly around 15.5, 14K around 13, 10K around 11.5, depending on what's mixed in. Every one of those numbers is dramatically higher than brass at about 8.5, steel at 7.8, or the zinc pot metal used in cheap castings at around 7. There is no overlap. If you get a real density figure, you have almost certainly answered the question.
You need a scale reading to 0.1 g, a glass of water, and thread:
- Weigh the dry piece. Suppose a pendant reads 12.9 g.
- Put a glass of water on the scale and press tare so it reads zero.
- Hang the piece by the thread so it's fully submerged but touching neither the sides nor the bottom. The scale now displays the weight of the water it displaced, which in grams is numerically its volume in cm³. Suppose it reads 0.99.
- Divide weight by volume: 12.9 ÷ 0.99 = 13.0 g/cm³. That's squarely in gold territory and rules out every base metal.
Had that same pendant read 1.52 cm³, the density would be 12.9 ÷ 1.52 = 8.5 g/cm³ — brass, and the conversation is over. Three things break the measurement: gemstones, hollow construction, and any air pocket that traps a bubble. Use density on solid, stone-free metal, tap the piece underwater to release bubbles, and take the number seriously when it's clean.
Test 4 — Put a Loupe on the Wear Points
Plating is thin — often a couple of microns, sometimes less than one — and it wears exactly where you'd expect. That makes visual inspection under magnification a legitimate diagnostic rather than a vibe check, and it's the test that most often catches an expensive-looking fake before anyone touches a chemical.
With a 10x loupe or a phone camera at full zoom, examine the places friction attacks:
- The underside of a ring shank, where it rubs against the adjacent finger
- The high points of any raised design or engraving
- The edges of a bracelet's links, and the inside curve where links pivot
- The back of a pendant, where it rides against a shirt
- Around the clasp and any jump rings
You are looking for a color change: a faint pink, gray, silvery, or greenish tone bleeding through the yellow. That's the base metal saying hello. Solid gold wears evenly and stays the same color all the way through, because it *is* the same all the way through. Also look at how the piece is put together — plated costume work often shows glue residue, seams, or crimped hollow tubing where genuine jewelry shows soldered joints.
A clean inspection doesn't prove authenticity, since a brand-new plated piece hasn't worn yet. But on any item with a few years of use, this test catches plating that a magnet never would.
Tests 5 and 6 — The Destructive Pair: Streak, Then Notch and Acid
The first four tests leave the piece exactly as they found it. The next two do not, so run them only after the free ones have failed to settle the matter, and never on something with antique, designer, or sentimental value.
Test 5, the ceramic streak. Drag the piece firmly across the unglazed back of a bathroom tile. Real gold leaves a gold-colored line. Plated base metal quickly gives up a gray or black streak as the coating gives way. Use a short stroke on a hidden area — this is sanding, and it leaves a fine mark.
Test 6, notch and acid. A jeweler's acid kit runs $15–25 and includes bottles graded for 10K, 14K, 18K, and 22K plus a touchstone. Here's the step that separates a real-versus-fake test from a purity test: file a small notch into a concealed spot first, deep enough to expose whatever lives under the surface. Then apply a drop of acid into the notch itself. Acid applied only to the outside of a plated piece reads the plating and reports "gold," which is how confident people end up wrong. Metal that fizzes, greens, or dissolves in the notch is base metal with a gold face. Metal that sits there unbothered is the real thing.
Treat the chemicals like the corrosives they are: nitrile gloves, eye protection, ventilation, a disposable surface, baking soda within reach, capped bottles stored away from children. The drops are tiny and the burns are not.
The Three Things People Call "Fake" — and Which Test Catches Each
"Fake" covers three genuinely different products, and knowing which one you're holding changes both the test that exposes it and whether the piece has any recoverable value at all.
| Type | What it actually is | Typical marks | Gold recoverable? | Test that catches it |
|---|---|---|---|---|
| Electroplated | A microns-thin gold layer over brass, steel, or silver | GP, GEP, HGE, or no mark | Effectively none | Wear points, streak, notch + acid |
| Gold-filled / rolled | A thick bonded gold layer, often 1/20 of total weight | GF, RGP, 1/20 12K GF | Yes, but refining minimums usually eat it | Notch + acid, weight vs. density |
| Counterfeit solid | Brass, tungsten, or pot metal with a faked karat stamp | A convincing 14K or 750 | None | Density, magnet, notch + acid |
| Gold over silver | Vermeil — real silver base, gold skin | 925 with gold color | Silver value only | Streak, XRF, acid |
The cruel case is gold-filled, because it *is* partly real gold and people reasonably expect to be paid for it. In practice, the layer on a typical 1/20 12K GF chain works out to a few percent of the item's weight in actual gold, and most refiners won't process small lots of it profitably. Our gold filled calculator works the arithmetic honestly. Vermeil is the sneakiest for home testers, because the silver underneath is a precious metal that behaves politely under several tests.
Test 7 — The Free XRF Scan That Ends Every Argument
Six tests in, most pieces have confessed. For the stubborn remainder, there's a professional-grade answer that costs nothing.
X-ray fluorescence reads a metal's actual elemental composition in a few seconds without marking it. Nearly every coin shop and bullion dealer owns an XRF analyzer, and if you walk in as a potential seller, the scan is almost always free — the dealer wants to know what your metal is even more than you do. You are not obligated to sell. Take the reading, thank them, and go compare offers. Ask to watch the screen; a dealer who won't show you the readout is telling you something.
XRF has one real limitation: it reads the surface, roughly the top layer of the metal, so a thick gold shell over a tungsten core can defeat it. That's a bullion-bar problem far more than a jewelry problem, and dealers handle it with ultrasonic thickness gauges that detect the boundary between two metals acoustically. For anything jewelry-shaped, XRF plus the density test you already ran is as close to certainty as a consumer needs.
One broader point about refusal, and it applies in both directions: if a buyer won't test your item in front of you, or a seller won't let you test theirs before money changes hands, that refusal is itself a result. The pressure tactics that ride along with it are catalogued in our guide to cash for gold scams, and the FTC's advice on selling gold jewelry makes the same recommendation: verify before you commit.
A Worked Case: The "Solid 18K" Estate-Sale Bracelet
Here's the whole funnel applied to one object, with the money at the end.
A link bracelet turns up at an estate sale for $60, stamped 750 on the clasp tag, weighing 24.6 g on a kitchen scale. The stamp claims 18K — 75% gold. If true, that's 24.6 × 0.75 = 18.45 g of pure gold. Assume gold at $4,400 per troy ounce; at 31.103 g to the troy ounce that's $141.46 per gram, so the melt value is 18.45 × 141.46 = $2,610. A $60 asking price for $2,610 of metal should trigger suspicion, not excitement.
Run the sequence. Stamp: present and plausible, crisply struck. Magnet: no response, so nothing steel inside. Density: 24.6 g dry, 1.59 cm³ displaced, giving 24.6 ÷ 1.59 = 15.5 g/cm³ — right where 18K belongs, and nowhere near brass. Wear points under the loupe: consistent yellow through the link pivots, no pink bleed. At this stage the piece has cleared every non-destructive gate, and you'd file a notch and acid-test it only if you still had doubts.
Now the part sellers skip. That $2,610 is melt value, not an offer. An online refiner at 70–90% of melt would land between $1,827 and $2,349. A local coin or bullion dealer at 65–85% would offer $1,697 to $2,219. A jewelry store at 50–75% would say $1,305 to $1,958, and a pawn shop at 40–60% would open somewhere around $1,044 to $1,566. Same verified bracelet, a spread of over $1,300 depending only on who you hand it to — which is why the authenticity question and the how much below spot to sell gold question are two halves of the same afternoon.
Frequently Asked Questions
What is the fastest way to know if gold is fake?
A strong neodymium magnet, in about four seconds. Gold is entirely non-magnetic, so any real pull from the body of the piece means steel or iron inside and the item is not solid gold. Test the main body rather than the clasp, since genuine chains often use steel clasp springs. The catch is that the fast test is also the weakest one: brass, copper, tungsten, and plated silver all ignore magnets completely, so a pass just means you keep going.
Can fake gold be distinguished from real gold at home?
Yes, with high confidence, provided you stack tests instead of relying on one. The combination that settles almost everything: a magnet to rule out steel, a water-displacement density measurement to rule out brass and pot metal, magnified inspection of wear points to catch plating, and acid applied inside a filed notch to check what's under the surface. Each covers another's blind spot. The rare failures are tungsten-cored bars and heavy gold over silver, and both are resolved by a free dealer scan.
Does real gold scratch easily?
Yes, and that surprises people. Pure gold is soft enough to mark with a fingernail, and even 14K is far softer than steel. But softness is a poor test, because lead, tin, and aluminum are also soft, while some plated pieces sit on a hard base that resists scratching entirely. Scratch behavior is a supporting observation at best. What matters is what the scratch reveals: if a scrape exposes a different color underneath, you've found plating, and that's the actual signal.
Is a piece with no hallmark automatically fake?
No. Plenty of genuine gold carries no stamp at all — most handmade jewelry, a lot of pre-1970s American work, pieces where the mark sat on a section that was later resized or repaired, and items where decades of wear simply polished it away. Absence of a mark means you have no claim to verify, not that the claim is false. Skip straight to physical testing: density plus a notch acid test will tell you what an unmarked piece is made of far more reliably than any stamp would have.
How do I test gold for authenticity at home without ruining it?
Use only the four non-destructive tests: read the hallmark under magnification, pass a strong magnet over the body of the piece, run a water-displacement density calculation, and inspect wear points with a loupe for base metal bleeding through. None of those touches the surface. Skip the ceramic streak and the file-and-acid step entirely for anything with antique, designer, or family value, and finish instead with a free XRF scan at a coin dealer, which reads composition without leaving a mark.
Will a gold-plated item pass an acid test?
It will, if you apply the acid to the outside — and that mistake is responsible for a lot of confident wrong answers. The plating is genuine gold, just microscopically thin, so acid on the surface reacts exactly as solid gold would. The fix is mechanical: file a small notch in a hidden area, deep enough to break through the coating, then drop the acid into the notch. Base metal underneath will fizz, discolor, or dissolve while the surrounding surface stays calm.
Does real gold float in water?
Never. Gold is one of the densest metals in everyday circulation at 19.32 g/cm³ pure, so it sinks immediately and decisively. The problem is that almost every metal sinks, including brass and steel, so the float test only catches plastic, resin, or hollow-cored novelties. The useful version of this test is quantitative rather than observational: weigh the piece, measure its displaced volume, divide, and compare the resulting density against the ranges for gold alloys and base metals.
What does 1/20 12K GF mean on a chain?
It marks a gold-filled item. The 12K describes the purity of the gold layer, and the 1/20 says that layer accounts for one-twentieth of the item's total weight. So on a 20 g chain, roughly 1 g of the piece is 12K metal — and 12K is half gold, so the actual pure gold content is around 0.5 g. That's real but small, and most refiners won't process modest gold-filled lots at all because processing costs exceed recovery. Treat gold-filled as jewelry, not as bullion.
Can a jeweler tell real gold from fake just by looking?
An experienced jeweler can often form a strong opinion from color, construction quality, wear pattern, and how the piece handles — and they're right far more often than a layperson would be. But they'll still reach for a tester before quoting money, because polished brass under warm shop lighting fools trained eyes routinely and modern counterfeits are made specifically to survive a glance. If someone offers to buy your piece based on visual inspection alone, ask them to scan it in front of you.
Is tungsten-filled gold a risk for jewelry?
Rarely. Tungsten's density of 19.25 g/cm³ sits within a fraction of a percent of pure gold's, which makes it the material of choice for faking large bullion bars where a single object carries tens of thousands of dollars and the interior is never seen. Jewelry is a poor target: it's thin, it gets cut and resized, and the shapes are hard to core convincingly. If you're buying bars rather than jewelry, buy from established dealers in sealed assay packaging and have secondhand bars ultrasonically checked.
My gold turned my finger green. Is it fake?
Probably not. Low-karat gold contains a lot of other metal by design — 10K is 41.67% gold and 14K is 58.33%, with copper, silver, nickel, and zinc making up the rest. Copper in particular reacts with sweat, lotion, and pool chlorine to leave a green or dark smudge on some people's skin. Meanwhile many counterfeit pieces are sealed under lacquer and mark nothing. Skin reaction tells you about alloy content and personal chemistry, not authenticity. Measure density instead.
Once I know it's real, how do I find out what it's worth?
Three inputs: weight in grams from a scale reading to 0.1 g, purity from the karat mark or a purity test, and the current gold price. Multiply weight by the purity decimal to get pure gold content, then multiply by the price per gram. Our karat gold calculator does it live. Then subtract reality: no buyer pays full melt, so expect roughly 70–90% from an online refiner down to 40–60% at a pawn shop, and get at least three quotes before agreeing to anything.
Are electronic gold testers worth buying?
For a household resolving one inheritance, no. The $25 units are fooled by thick plating and get confused by white gold alloys, while the models that dealers actually trust start around $150 — more than most testing questions are worth. They earn their keep for someone buying estate lots regularly, where ten seconds per piece with no chemicals is genuinely valuable. If you do buy one, calibrate it against a known-genuine piece first and treat any single reading as a hint rather than a verdict.
Should I test a coin the same way I test jewelry?
Partly. Magnet and density both apply and both work well, since bullion coins have published weights and purities that a fake will struggle to match on both measures at once — weigh the coin and compare to the official spec before anything else. But skip streak tests, notches, and acid entirely: damaging a coin destroys any collector premium, which can be a large multiple of its metal content. Take questionable coins to a dealer for a free scan, and see melt value of gold coins for the specs to check against.
Written by Sukie Gao
Sukie Gao holds a master's degree from a business school, where she picked up the markets-and-pricing toolkit she now applies to the consumer gold trade. She created Gold Calculator Hub to give people an independent, data-driven way to find out what their gold is really worth.
Published September 3, 2026