THC limits: which number applies where

A rule written for growers

A THC limit is a maximum permitted amount of delta-9-THC, written into law, for one clearly defined object. The object matters more than the figure. And the figure nearly everyone has heard, 0.3%, was written for a plant growing in a field.

It lives in EU agricultural law. Its function there is variety cultivation eligibility under the Common Agricultural Policy, which is the formal way of saying it decides which hemp varieties a farmer may put in the ground and still have the crop count as hemp. The audience is growers. The sample comes off the crop. The result lands in an agricultural file.

Picture the paperwork behind it. A variety register, a certified seed lot, an inspector in a field with a sample bag, a laboratory result on plant material. Nobody in that chain is holding a bottle, a jar or a pack of shelled seeds. The scope is the field, not the shelf.

The threshold also exists in two versions. Until 2023 it read 0.2%. It was raised to 0.3%. Both numbers are still in circulation, and both get repeated as though they governed finished goods.

Here is the part that usually goes missing. The cultivation threshold is silent on the delta-9-THC content of an extract, a cream or a food. Processing, dilution and formulation all happen after harvest, and the field rule does not travel with the material through any of them. A compliant variety is where a supply chain starts. It is not a statement about what sits in a 10 ml pipette bottle.

Feed has its own figure. Hemp material intended for feed use is handled under 0.2%. Same plant, different destination, different number, different paperwork.

None of this makes 0.3% wrong. It makes it narrow. Read against a field, it answers the exact question it was built for. Read against a finished product, it answers a question that law never asked.

Converting percent into mg/kg

Percent and mg/kg carry the same information in different clothes, and the conversion is plain arithmetic. 0.3% is 0.3 g per 100 g. That is 3,000 mg/kg. Keep that number nearby, because every comparison below runs through it.

Food has its own limits, in mg/kg, split by category. Commission Regulation (EU) 2022/1393 sets 3.0 mg/kg delta-9-THC for hemp seeds and dry hemp-seed foods, and 7.5 mg/kg for hemp seed oil. Two products from one plant, two values, and a regulation identifier that hardly anyone recognises on sight.

Cosmetics are assessed on a separate track. The Scientific Committee on Consumer Safety, in the opinion carrying the identifier SCCS/1685/25, works with a THC impurity level of 0.00025%. Converted, that is 2.5 mg/kg. It is written as an impurity figure, which tells you what kind of number it is meant to be.

Now put them in one column, all in mg/kg. Cultivation: 3,000. Hemp-seed food: 3.0. Hemp seed oil: 7.5. Cosmetic THC impurity: 2.5. Feed sits under 0.2%, so 2,000. The field figure lands at 1,000 times the hemp-seed food limit, 400 times the seed-oil limit, and roughly 1,200 times the cosmetic impurity figure.

So a phrase like "under the 0.3% limit" can be perfectly accurate about a crop and still say nothing about the category a product belongs to. The product limits are fragmented by category by design. Each has its own object and its own text.

There is one more small tax on mg/kg. A percentage is self-contained. A milligram-per-kilo value needs a second quantity, the mass of the product, before it means anything at all about the container in your hand.

Reading a batch certificate

Take a number from our own paperwork. A batch certificate for one of our stronger oils reports 0.03% delta-9-THC. Convert it: 0.03 g per 100 g, so 300 mg/kg.

Against the cultivation figure, 300 next to 3,000, that is one-tenth. It reads comfortable. Against the hemp-seed food limit of 3.0 mg/kg it is 100 times higher. Against the 7.5 mg/kg for hemp seed oil, 40 times higher. Same measured value, three different verdicts, depending entirely on which yardstick you hold it against.

The honest reading is the boring one. Compare a measured content with the limit for the product category in front of you, not with the limit for a field.

Then there is the other phrase certificates use. "Not detected" means delta-9-THC was looked for in that batch and not found above the method's reporting threshold. It is a statement about a method and a batch. It is not proof of compliance with 0.3%, because 0.3% is not the finished-product question.

Six things worth checking on the paperwork:

• Which analyte was measured: delta-9-THC on its own, or a total THC value that folds in the acidic form.
• The unit printed on the report, percent or mg/kg, and the conversion between the two (0.03% = 300 mg/kg).
• The reporting threshold of the method, so a "not detected" line has a real value standing behind it.
• The batch number on the certificate, matched against the batch number on the bottle.
• The category the product belongs to, since food, feed and cosmetic figures sit in separate texts with separate numbers.
• The mass of the product, whenever the applicable limit is written in mg/kg.

Why the field figure stuck

0.3% made a journey no product limit has managed: out of agriculture and into marketing copy. It has two advantages the product limits lack. It looks like one number covering everything, and it is a percentage, so it needs no second quantity to sound meaningful.

The product limits arrive with the opposite profile. They are fragmented by category, they come in mg/kg, and they travel under identifiers such as 2022/1393 or SCCS/1685/25 that few people recognise at a glance. A number that needs a regulation identifier and a product mass before it says anything will always lose ground to a number that fits in four characters.

The two versions of the farming figure add to the mess. 0.2% before 2023, 0.3% after, both still repeated, both offered as rival product limits by people with no particular reason to know the difference.

And then there is paraphrase drift, which you can watch happen in real time. A certificate states that THC was not detectable. Someone summarises that as "below 0.3%". The summary sounds tidier and more official, and in a single step the farming figure has been reinstated as the benchmark for a bottle.

Our house rule since 2014

We have been working with cannabinoids since 2014, and the rule we settled on is deliberately plain. Publish the measured number. Name the yardstick it is being compared against. Leave the applicable comparison to the reader, who knows which category the product belongs to and can convert a percentage into mg/kg in one step. Every product gets its own batch certificate, so the value on the page comes from a specific batch rather than from a range. If a figure looks unfamiliar, the certificate is the place to check it. Claims are easy. Measured values are better.