The Safety Limit That Isn't A Safety Limit

In December 2024 California's Department of Pesticide Regulation published a sixty-six page memorandum recommending revised pesticide action levels. Page five explains how the limits for inhalable products were derived.

There was, the department noted, "currently limited data specifically characterizing pesticide inhalation exposure". So it adopted Guidance Residue Levels established for tobacco — levels which, in its own words, "reflect the highest acceptable residues resulting from agricultural practices, detection limits, and physical and chemical properties".

Read that list again. Agricultural practice. Detection limits. Physical chemistry. Human health is not on it. Such a level answers the question *what residue is achievable if a grower follows the label* — not *what is safe to inhale*.

Where no tobacco guidance level existed, the department fell back on a default of 0.1 micrograms per gram. Its memorandum is candid about the provenance: "This is the pesticide residue level for tobacco that triggers the US EPA to require pyrolysis testing."

In its native context that figure is not a safety threshold at all. It is the point at which a regulator says *go and find out what happens when this burns*. Transplanted into a different framework it became the thing you are measured against — and the study it exists to trigger is never performed.

The same document, one page earlier, sets out precisely why inhalation deserves separate treatment: the route "carries unique risks because heating or burning pesticides during smoking or vaping products may release toxic pyrolysis or thermal decomposition products such as hydrogen cyanide or chlorine compounds, which can directly enter the bloodstream from the lungs". The department identifies the hazard, records that data characterising it are limited, and then reaches for tobacco farming figures.

The same regulator does derive genuinely health-based limits — but only for products that are eaten. That calculation divides a reference dose by a consumption rate, and the reference dose is defined in the memorandum as "an estimate of a daily oral chemical exposure".

When the department updated its methodology in August 2025, incorporating real consumption survey data for the first time, it recorded plainly that "no additional updates for action levels for inhalable … products will be made at this time". The refinement went to edible products. The inhalable numbers remain the borrowed ones.

Washington State's regulator reached the same conclusion in a September 2025 research brief and put it more bluntly: no pesticides "have been tested for human toxicity for the different ways in which products are used (e.g., inhaled). This has resulted in the lists of pesticides required to be tested in a lab to be primarily based on the ability to detect pesticides and not on human health and safety."

The same brief adds a warning for anyone tempted to lean on tobacco as a proxy: users of inhaled plant products "inhale larger volumes of smoke and hold inhalations for longer in the lungs, which greatly limits comparisons to tobacco users".

In March 2026 the same team put it into the peer-reviewed literature. Writing in *Clinical Therapeutics*, Watson and colleagues concluded that action levels "are not based on human health risk" and are instead "derived from analytical detection limits". Their paper singles out patients — people using such products frequently, often alongside other health conditions — as the group most exposed to the consequences.

The Australian standard for inhaled botanical medicines specifies pesticides in a single line of a schedule. The parameter is "Pesticides", the test method is the European Pharmacopoeia's chapter 2.8.13, and the limit is "not more than the limits specified in Ph Eur 2.8.13". The Australian instrument sets no pesticide figures of its own. It delegates entirely.

Follow the delegation and something interesting happens. The European chapter is admirably clear that growth regulators fall within its scope — its definition expressly "includes substances intended for use as growth-regulators, defoliants or desiccants". But its table of named substances, the list a laboratory actually works from, consists of organochlorines, organophosphates, pyrethroids and dithiocarbamates. The common growth regulators are not on it.

For anything unlisted, the chapter engages only where a substance's presence is "suspected for any reason", and then sets a limit calculated from an oral acceptable daily intake.

So the Australian position is: growth regulators are covered in principle, named nowhere, tested for only on suspicion, and — where a number is derived at all — measured against a figure built for swallowing. Meanwhile the compounds themselves are ordinary registered agricultural chemicals here, one of them with thirty-five commercial products approved for apples, mangoes, avocados and stone fruit.

None of this establishes that residues on any given product are dangerous. It establishes that nobody has demonstrated they are safe by the route that matters, and that the numbers implying otherwise were built to answer a different question.

A limit derived from what a sprayer can achieve, or from what an instrument can see, tells you about sprayers and instruments. It does not tell you about lungs.

The defensible conclusion is therefore not *the residue is under the limit, so it is fine*. It is *the limit was never a health limit, so the only target that requires no assumptions is none at all*. Precaution is not timidity when the alternative is inference from a number that was never about safety.