TEYING BUYER GUIDE · UPDATED OCTOBER 23, 2026
REACH Metal Limits for EU-Bound Jewelry: Parts, Not Pieces
Lead tops out at 0.05% by weight, cadmium at 0.01%, nickel release at 0.5 µg/cm²/week. What each REACH limit measures, and where test reports fail.

A jewelry order rarely fails at a border because nobody read the regulation. It fails because a two-cent jump ring, or the solder under a stone, was never on the test plan. EU metal limits are written against individual parts and individual materials, which is why the headline numbers are easy to quote and stubborn to satisfy across a full bill of materials.
The three numbers, and what each one measures
Lead sits under REACH Annex XVII entry 63: no more than 0.05% by weight (500 mg/kg) in any individual part of a jewelry article. Cadmium sits under entry 23: no more than 0.01% by weight (100 mg/kg) in metal and plastic materials. Nickel works differently. Entry 27 caps release at 0.5 µg/cm²/week for prolonged skin contact, with 0.2 for post assemblies in pierced body parts.
Two of the three are content limits read from the material. The third is a release rate read off the finished article after simulated wear. All three are enforced on the piece that reaches the market, not on the drawing that was approved.
| Limit | Annex XVII entry | Measured as | Applies to |
|---|---|---|---|
| Lead | 63 | Share of weight | Each individual part |
| Cadmium | 23 | Share of weight | Each metal or plastic material |
| Nickel | 27 | Release per area per week | The finished article, in skin contact |
Treat this as general guidance as of October 2026, not legal advice. The entry text on EUR-Lex or ECHA is the version that binds, not a marketplace help page and not this table.
Why "per part" drives the whole test plan
Entry 63 is drafted part by part. A clasp that passes says nothing about a pendant whose solder or casting alloy carries lead, and one failing component is enough to pull the product. Entry 23 is five times stricter than lead on paper, and in practice it is the most common reason a low-cost alloy jewelry report comes back red.
A caution on entry numbers, because it costs people real money: several compliance pages still in circulation swap the numbers for lead, cadmium and nickel. Read the wording beside the number, including here.
Where the entry text differs from what buyers assume
Entry 63 lists exemptions. Crystal glass as defined in Directive 69/493/EEC is out. Internal watch components a consumer cannot reach are out. Non-synthetic or reconstructed precious and semiprecious stones are out unless they have been treated with lead. Enamels produced by fusion at 500 °C or above are out. Jewelry placed on the market before 9 October 2013 also sits outside the restriction, which is no help to anyone producing new stock.
The figure most often borrowed from the wrong place is 0.5%. An allowance for brass alloys carrying lead up to 0.5% by weight does exist, but it belongs to a different paragraph of the annex covering articles children may put in their mouths, not to the jewelry restriction in entry 63 paragraph 1. Free-machining brass at a fraction of a percent of lead is still roughly ten times over the jewelry limit. Ask for the alloy designation and a lead result per metal part instead of treating "brass" as an answer.
What the enforcement record actually looks like
Public Safety Gate data across December 2025 and January 2026 shows a cluster of jewelry notifications for excess cadmium, nickel and lead, with Chinese-origin products prominent. The actions recorded are not website warnings: imports rejected at the border, stock destroyed, listings pulled from marketplaces, consumer warnings issued, recalls run.
Two alerts show the range. A piercing jewelry alert notified by Finland in January 2026 cited cadmium at 87.9% by weight in the article, with withdrawal from end users effective 8 December 2025. A silver bracelet alert cited nickel release up to 24.63 µg/cm²/week against the 0.5 limit, cadmium up to 0.017% and lead up to 2.3% by weight, with the import rejected at the border on 12 December 2025. Recall reporting for the first quarter of 2026 logged roughly 50 jewelry notifications, cadmium leading at about half, nickel next at around 30 percent and lead at around 18 percent.
Read that as category enforcement rather than a case against one seller. The buyer-side response is to know which SKU in your range carries the highest exposure and to hold a report that covers it. Safety Gate publishes weekly; searching your product category once a quarter is a reasonable habit.
Reading a metals report field by field
| Field | What good looks like | What should make you ask again |
|---|---|---|
| Identity | Report number, issue date, sample name, photo, SKU and batch | "Earrings, gold plated" with no SKU, or a report from an earlier production run |
| Laboratory | Name, address and ISO/IEC 17025 accreditation, with the methods inside the accredited scope | A logo with no accreditation number you can check on the body's public register |
| Method | Acid digestion followed by ICP-OES or ICP-MS, with limits of quantification stated | A result with no method and no stated LOQ |
| Nickel specifics | EN 1811:2023 cited, plus whether EN 12472:2020 wear simulation was applied | A nickel figure with no wear step described |
| Coverage | Post, body, plating on both, solder and any stone setting in skin contact | One result for the whole item treated as a single sample |
| Margin | Values with units, sitting next to the limit that applies | A cadmium result at 95 mg/kg read as "pass" with no thought about variation |
| Change control | A new plating supplier, alloy batch, solder or switched finding triggers a retest | Silence on who pays for the retest and how long it takes |
A pass is evidence about one sample. It is not a GPSR technical file, and it does not move responsibility from the factory to the laboratory.
Five lines worth writing into the purchase order
- Name the prohibited metals and write the limit values out, instead of accepting "REACH compliant" as a sentence.
- Require an alloy designation for every metal part, and the cadmium status of the plating process rather than the color of the finish.
- Cover the small parts explicitly: solder, jump rings, ear posts, clasps and stone settings, where a compliant body usually meets a non-compliant component.
- Test per part on the first run of a new design, then per batch on a schedule both sides signed.
- Keep the sample the report was run on, sealed and labeled, so a later dispute has a physical reference.
How TEYING handles the material question
We work in brass and 925 sterling silver, with a four-layer plating system running 0.5 to 2.5 µm depending on the wear target. Sampling runs 14 to 18 days, and production starts at 60 pieces per design. Lead-free, nickel-free and cadmium-related requirements can be raised before production begins, and the material and plating route is declared per SKU on the quotation rather than described in adjectives.
Applicable testing documentation is confirmed by material, destination market and order scope. If a design changes after the report date, we say so and you decide whether the batch needs new testing. That is the honest limit of what a factory can promise, and a claim of standing certification broader than a test report deserves a question wherever it comes from.
If you are building an EU-bound order and want the material and plating detail in writing before sampling, send the design and quantity through the TEYING quote form.
FAQ
Which of the three limits fails most often on brass jewelry?
Cadmium, by a wide margin. Entry 23 allows 0.01% in metal and plastic, five times tighter than the lead ceiling, and low-cost alloy stock plus bright plating processes are where the number usually breaks. Cadmium also tends to arrive inside the base material rather than on the surface, so a plating change rarely rescues a failing casting.
Does one passing report cover a reorder of the same design?
Only if nothing moved. A new alloy batch, a new solder, a different plating supplier, a thicker deposit or a changed stone can all shift the result while the drawing stays identical. Treat the report as tied to a production run and a sample, and retest whenever one of those inputs changes, or agree a batch schedule in advance.
Why is a cadmium failure harder to fix than a lead failure?
Lead often enters through solder or a coating, so changing that one input can solve it. Cadmium more commonly sits in the alloy itself, which means a new material lot, a different casting source, or a different base metal. The reliable sequence in both cases is the same: find the part that fails, change the input at that point, then retest the finished article.
Can I rely on the 0.5% lead allowance for brass alloys?
Not for jewelry. That allowance sits in a different paragraph of Annex XVII, covering articles children may put in their mouths, not in the jewelry restriction under entry 63 paragraph 1. A free-machining brass carrying lead at a fraction of a percent is still far above 0.05% and will fail the article limit.
How do I test an earring that has plating, solder and a set stone?
Split it into the parts that touch skin and test each: the post, the body, the plating on both, the solder joint and the setting. One combined sample for the whole earring averages the results and hides the part that failed. Ask the laboratory to name each part in the report so every result maps back to your bill of materials.
EU-bound order with a metals requirement?
Send the material and plating direction; lead, nickel and cadmium requirements are discussed before production.
Request a Factory Quote