1How to read a maximum level

Most producers look these levels up because someone has asked them for a test report. A retailer asks before listing a product, an importer needs the documentation to clear customs, a marketplace requires a certificate of analysis from an accredited laboratory, or an inspector takes a sample from the shelf. Some test before anyone asks. In each case they need to know two things: which level applies to their product, and whether the analysis they are buying can measure it.

Regulation (EU) 2023/915 replaced Regulation (EC) No 1881/2006 in May 2023. Annex I of the regulation lists the maximum levels for contaminants in food, and section 3 of that annex covers metals.

Each row gives the product, the element, the level in mg/kg and a remarks column. The product name is a legal definition and often refers to another regulation, so a product that looks as though it belongs in a row may not. The remarks column sets the conditions and the exclusions.

The levels apply to the wet weight of the product as sold, unless a row says otherwise. The regulation has been amended ten times since 2023, so the version to read is the consolidated text. The current version is dated 8 October 2025 and carries the identifier 02023R0915 — EN — 08.10.2025 — 006.001. Two of the amendments matter here: nickel was added in 2024, and the arsenic section was rewritten in September 2025.

2Supplements

The regulation sets four levels for food supplements.

The cadmium level depends on what the supplement is made of. The general level is 1.0 mg/kg. It rises to 3.0 mg/kg only if the supplement consists of at least 80 % dried seaweed, seaweed-derived products or dried bivalve molluscs. A supplement containing 70 % seaweed falls under the general level and is judged against 1.0 mg/kg.

Where the metals come from

Cadmium enters through the plant. Seaweed, cocoa and several leafy crops take it up from the soil as they grow, which is why the regulation allows a higher level in seaweed-based supplements. Lead usually enters during handling: soil on a root crop, contact with equipment, or a glazed container. It has also been added deliberately — lead chromate has been used to deepen the colour of turmeric. Mercury accumulates in the marine food chain and reaches supplements through ingredients such as fish oil and algae.

What the regulation does not say

The regulation sets no arsenic level for supplements. Section 3.4 covers cereals and rice, infant formulae, baby food, fruit juices, fish, crustaceans, bivalve molluscs, cephalopods and salt, and supplements are not on that list. We measure arsenic in supplements and report the concentration, but there is no EU level to compare it with. Where a supplement is made from rice, algae, seaweed or seafood, the level that applies belongs to that ingredient and comes from its own row.

The regulation sets no nickel level for supplements either. Nickel was added to Annex I in 2024 for cereals, juices and several other foods, but not for supplements.

The regulation also does not define a food supplement. Article 1 gives seven definitions and none of them covers the term, and Directive 2002/46/EC, which does define it, is not cited anywhere in the text.

3Spices and botanicals

For dried spices the regulation sets levels for lead only. There are six rows, divided by the part of the plant the spice comes from, and the levels differ widely between them: cinnamon is allowed more than three times the lead permitted in peppercorn.

The regulation names the six categories but does not say which spice belongs to which. A footnote at the head of Annex I refers to Annex I of Regulation (EC) No 396/2005, category 0800000, as replaced by Regulation (EU) 2018/62. The right-hand column above comes from that regulation, not from 2023/915.

That cross-reference leaves one gap. Category 0800000 has a seventh entry, aril spices, which covers mace, and the regulation sets no lead level for it. Nutmeg is a seed spice at 0.90 mg/kg; the aril that grows around the same seed has no level at all.

Cadmium, mercury and arsenic in spices

The regulation sets no level for cadmium, mercury or arsenic in spices. The cadmium section runs from 3.2.1 to 3.2.21 without a spice row. Mercury covers only fishery products, bivalve molluscs, supplements and salt. Arsenic covers the categories listed above, and nickel does not include spices. A spice tested for all four elements therefore returns one result with a legal threshold behind it and three without, and the report should say which is which.

4Wine, beer, cider and spirits

Of these four drinks only wine has a maximum level, it applies to lead alone, and it depends on the vintage. The level has been lowered twice, and each reduction applies from a harvest year rather than from a calendar date.

The remarks column of row 3.1.20 states that the row includes semi-sparkling and sparkling wines and excludes liqueur wine and wine of not less than 15 % vol. Liqueur wine has its own row, and only from the 2022 harvest onwards, so an older bottle has no level. Wine of 15 % vol or more falls outside row 3.1.20 and no other row covers it: a fortified wine can be measured, but the result cannot be compared with an EU level.

Beer and spirits

The regulation sets no level for lead, cadmium, mercury, arsenic or nickel in beer or in spirits. The word beer appears twice in Annex I, both times in an exemption concerning the cereal used to brew it, and spirit drinks appears once, under patulin, which is a mycotoxin. Both can be analysed and the results are valid measurements, but there is no level to compare them with.

The method for lead in wine

For lead in wine the regulation does not leave the method open. Article 80(5) of Regulation (EU) No 1308/2013 refers the analysis to the methods of the International Organisation of Vine and Wine, and the sampling and analysis regulation follows that reference.

5Juices and nectars

Juices carry more rows than the other three matrices, and the most recent of them: nickel applies from 1 July 2025.

A juice made only from berries and other small fruits is judged at 0.05 mg/kg of lead. Every other juice, including any mixture, is judged at 0.03 mg/kg, so a blend of berries and apple falls under the stricter level.

Vegetable juices appear in one row only, the nickel row. The lead and inorganic arsenic rows are written for fruit juices and nectars, so a tomato or beetroot juice can be measured for all five elements and four of the results have no threshold behind them.

For a concentrate the level applies to the reconstituted juice. Row 3.1.19 says so in its remarks and row 3.4.4 says so in the product name. The dilution factor therefore forms part of the analysis, and the customer has to supply it.

6What a result can and cannot establish

The levels above are thresholds. Whether an analysis can establish that a product meets one depends on the method, and the same two regulations set out how.

Total arsenic and inorganic arsenic are different measurements

Arsenic in food occurs in several forms. The inorganic forms, arsenite and arsenate, are the toxic ones. The organic forms, mainly arsenobetaine, pass through the body without harm. A fish can contain a large amount of arsenic and still be safe, because almost all of it is organic.

The EU therefore sets its levels for inorganic arsenic rather than for the total. Routine ICP-MS measures the total: every form at once. Measuring the inorganic fraction on its own requires chromatography coupled to the ICP-MS, and that is a separate analysis.

The inorganic fraction is part of the total and cannot exceed it. A total below the level therefore establishes compliance without measuring any species at all. Regulation (EC) No 333/2007 sets this out as the rule for official control:

“Methods for analysis for total arsenic are appropriate for screening purpose for control on inorganic arsenic levels. If the total arsenic concentration is below the maximum level for inorganic arsenic, no further testing is required and the sample is considered to be compliant…”

A total at or above the level does not settle the question, and the same rule requires follow-up testing.

The reporting limit the regulation requires

A result reported as less than the reporting limit is only useful if that limit lies well below the maximum level. Regulation 333/2007 makes this a requirement and sets it relative to each level. For lead, cadmium and mercury, where the maximum level is 0.1 mg/kg or above, the limit of quantification must be no more than one fifth of it. For arsenic it must be no more than two thirds, and no more than the level itself where that level is 0.03 mg/kg or below. For nickel it must be no more than the level where that level is 0.3 mg/kg or below.

The table below applies those criteria to the rows above. The third column gives what the regulation requires; the fourth gives what our method achieves.

Two rows meet the criterion exactly rather than falling below it: mercury in capsules, and inorganic arsenic in juice. Both satisfy the requirement, but neither leaves room to spare.

What the result applies to

A result applies to the sample submitted and to nothing else. Official control does not test a single retail unit. It takes incremental samples across the lot and combines them into an aggregate sample, normally one kilogram, although for supplements, dried spices and herbs, dried fungi, algae and lichen the regulation reduces that aggregate sample to 100 grams. Testing one pack tells you about that pack. Whether that pack represents the batch depends on how the sample was taken.

Measurement uncertainty is part of the decision

Every result carries a measurement uncertainty, reported as x ± U with a coverage factor of 2. A result of 0.11 mg/kg against a level of 0.10 mg/kg is not automatically a failure. The regulation rejects a lot only where the result exceeds the maximum level beyond reasonable doubt, taking the expanded uncertainty into account, and accepts it where the result does not exceed the level on the same basis.

The regulation states that principle but does not give the calculation. The familiar operation, subtracting U from the result before comparing it, comes from a Commission report cited in the annex rather than from the legal text. A result that lies within its own uncertainty of the level is therefore neither a pass nor a failure, and the report should say so.

Dried, diluted, processed, compound

Where Annex I has no row for the form a product is actually in — a powder rather than a liquid, an extract rather than a plant, a mixture of several ingredients — Article 3 applies. It requires the concentration or dilution caused by processing, the proportions of the ingredients and the limit of quantification to be taken into account, and it places the burden of justifying the factor on the food business operator. Concentrated juice is an exception, because the rows state that the level refers to the reconstituted juice.