A wedding cake topped with live roses. A glass of prosecco with a petal floating in it. A salad garnished with a chrysanthemum pulled from yesterday’s bouquet. It looks expensive, it costs almost nothing, and social media is full of it.
In Belgium, researchers bought 90 bouquets from 50 flower shops and sent them to an accredited lab. They found 107 pesticide active substances. Roughly ten per bouquet. The dirtiest bouquet carried 97 mg of residues per kilogram.
Here is where that comes from, why rinsing does not fix it, and what to put on the cake instead.
Why cut flowers have no limits at all
Food has maximum residue limits. Every pairing of a substance and a crop has a number attached, plus a pre-harvest interval: a required gap between the last spray and the harvest, long enough for the residue to drop below the limit.
A cut flower is not food. So it has neither a limit nor an interval. The Belgian authors say it plainly.
The lack of maximum residue limits (MRL) for flowers explains that, unlike other crops which are harvested for consumption, there is no restriction on the use of pesticides and cut flowers are often treated regularly up to harvesting or even after harvest.
Those last four words answer the question in the title. The rule is not “spray until harvest, then wait.” There is no waiting, and treatment continues after the stem is cut.
The grower’s logic is easy to follow. A rose has to travel from Kenya or Ecuador without a single blemish, because a blemish means an unsellable stem. Pest pressure in a greenhouse is high, the product is valuable, and nobody shows up to test for residues. Under those conditions you spray as much as the appearance requires.
What the bouquets actually carried
Belgium, 2016. Fifty bouquets of roses, 20 of gerberas, 20 of chrysanthemums. The analysis used the same multiresidue method applied to food, with a limit of quantification of 0.01 mg/kg.
| Flower | Substances per sample | Total residues |
|---|---|---|
| Roses | 14 on average | 26 mg/kg |
| All 90 bouquets | about 10 on average | up to 97 mg/kg in the worst case |
The single highest concentration was dodemorph, a fungicide used against powdery mildew on roses, at 41.9 mg/kg. Propamocarb followed at 35.4 and procymidone at 35.3. Why powdery mildew is so hard to stop without products like these, and which greenhouse conditions set it off, is explained in the wiki on powdery mildew and how to prevent it.
By chemical family the picture is 12 organophosphates, 8 pyrethroids, and 7 carbamates. All three groups act on the insect nervous system, and the mechanism in a mammal is the same one, with dose as the difference. The authors name the acutely toxic members separately: acephate, methiocarb, monocrotophos, methomyl, deltamethrin.
There is another column people rarely think about. Substances classified as sensitizers, meaning capable of triggering allergy, came to 16 in roses, 12 in gerberas, and 11 in chrysanthemums.
None of this is new. Back in 1977, Miami tested every flower shipment arriving over three consecutive days: 18 of 105 bouquets carried more than 5 ppm of residues, and three samples exceeded 400 ppm.
One odd sample proves nothing, so here is a second
Austria and Germany, 237 cut flowers bought from 22 garden centers and retailers between 2011 and 2021, screened for 646 substances.
- Residues turned up in 97% of cut flowers
- 11 substances per flower on average, twice the load found on potted plants
- 72% carried substances classified as harmful to human health
- 85% carried persistent or very persistent substances, the kind with a half-life of 100 days and up
Two independent samples, different countries, different labs, eight years apart. The numbers line up: about a dozen substances per bouquet is the normal state of affairs.
How much comes off on your hands
This is the part that matters for eating. If a petal is going on a cake, the question is not how much is on it in total, but how much leaves it.
Twenty Belgian florists worked two to three hours wearing cotton gloves. The gloves went to the lab.
They carried 111 active substances, averaging 37 per glove sample. The most frequent, present in 90% of samples: benomyl together with its metabolite carbendazim, boscalid, iprodione, clofentezine, lufenuron, flonicamid.
When the systemic dose was modeled, four substances exceeded the acceptable operator exposure level in the worst-case scenario. Clofentezine exceeded it fourfold, reaching 393%. And urine samples from Belgian florists contained an average of eight residues and metabolites each.
So these substances do not stay welded to the petal. They transfer on contact, cross the skin, and show up in urine. A petal sitting on buttercream for half a day does the same thing, into the buttercream.
The second wave: what happens after cutting
Pesticides are half the story. The other half starts once the stem is cut and somebody has to get it to the shop alive.
| Treatment | Purpose | Typical rate |
|---|---|---|
| Silver thiosulfate, silver nitrate | block ethylene, delay wilting | up to 1500 ppm silver |
| 8-hydroxyquinoline (8-HQC, 8-HQS) | bactericide in the vase solution | 100-500 mg/L |
| Benomyl, thiabendazole | fungicides, applied by dipping the flower head | 200 mg/L |
| Benzyladenine | growth regulator, spray or dip | 200 mg/L |
| Sodium hypochlorite, aluminum sulfate, boric acid | water sanitation | per protocol |
| Citric acid, sugar | acidify to pH 3.5 and feed the stem | sugar from 2% to 20% |
| Wax coatings (Wilt Pruf, Folicote) | cut water loss | dilutions of 1:4 to 1:40 |
Look at the benomyl row. It goes on by dipping the flower head, after harvest. Benomyl is not approved in the European Union, and its metabolite carbendazim is classified as toxic to reproduction. That exact pair showed up in 90% of the florists’ glove samples.
The little sachet of flower food that comes with a bouquet is the same logic in miniature: sugar, an acidifier, and a biocide. The biocide is what keeps the vase water from going cloudy, and it is also what makes vase water what it is.
Why washing does not help
The first instinct of any sensible person is to rinse it under the tap. That fails for three reasons.
- They are fat soluble. The Belgian authors note specifically that most substances applied to flowers are persistent, transfer by touch, and dissolve in fats. Water does not lift those compounds off a surface. Buttercream lifts them very well.
- Some of it is inside. Neonicotinoids and a number of fungicides are systemic: the plant takes them up through the roots and distributes them through its tissue. Whatever sits inside the petal cannot be rinsed off by anything, in principle.
- A petal cannot survive washing. You can scrub a cucumber with a brush. A rose petal disintegrates long before you have removed anything.
The logic matches what happens with vegetable wash sprays: rinsing removes what sits on top and dissolves in water, and does nothing about the rest.
Putting 97 mg/kg in context
The number means nothing without a yardstick. Take clofentezine, the substance that exceeded the operator exposure level fourfold. Its EU food limits look like this:
| Food | Limit, mg/kg |
|---|---|
| Table grapes | 0.02 |
| Cucumbers | 0.2 |
| Tomatoes | 0.3 |
| Apples, pears | 0.5 |
| Strawberries | 2.0 |
Strawberries may legally carry 2 mg/kg of clofentezine, the highest figure in that table. Roses averaged 26 mg/kg in total, and one bouquet hit 97. The order of magnitude differs for a simple reason: somebody calculated a number for strawberries, and nobody was ever asked to calculate one for roses.
It is worth seeing what is permitted even in organic production, and at what rates. Limits exist there, they are strict, and the presence of a limit is exactly what separates food from decor.
What to put on the cake instead
- Food-grade edible flowers. Grown as food, under food rules, with pre-harvest intervals. Sold by the same suppliers who sell restaurant microgreens, not by the florist on the corner
- Your own windowsill. Nasturtium, calendula, marigold, and viola grow fast and need nothing but water. Watering edible flowers works the same way it does for greens. For what blooms outdoors and when, see the edible flower bloom calendar
- Culinary herbs in bloom. Basil, thyme, cilantro, and arugula all throw small flowers. They are edible by definition and look no worse on a plate
- If the flower is decor rather than food, keep a barrier between it and the product: a disc, a film, a layer of fondant. And remove it before anyone cuts the cake
There is no point asking a florist whether the flowers are edible. They will honestly say they do not know, and they will be right: no residue certificate travels with the stems, and there is no way for them to obtain one. That information simply is not passed along the chain. And if you pick flowers yourself, first make sure the plant has no poisonous lookalike: see toxic lookalikes of edible flowers and greens.
Which flowers actually go on a cake
For baking, the Royal Horticultural Society (RHS) names rose, sweet violet and viola, primrose, English daisy, borage, lavender, and dianthus. Rose, violet, borage, and primrose get crystallized, and daisy, borage, and primrose also go on fresh. Lavender and dianthus more often end up as flavored sugar. On large flowers such as roses, use only the petals and discard the stamens, pistil, and calyx.
A pot from the garden center is not a food ingredient yet. If the label does not say it is for culinary use, the RHS advises growing the plant on for at least three months and harvesting only the flowers that open after that.
In 2023 the UK Food Standards Agency (FSA) went through this topic in a reply to a cake business. A few of its points are easy to miss:
- Sugar does not detoxify. Crystallizing removes some of the problems of fresh flowers, but a poisonous plant in sugar is still poisonous
- Stems in the cake need a barrier. Florist wire will not do; use a food-safe cake pick, and make sure sap from the cut end cannot run onto the frosting
- Drying counts as food contact too. Sand and silica gel used for drying must be food grade. A flower soaked in glycerin or coated in paraffin wax becomes inedible, whatever it was to begin with
- A lookalike is a different plant. Jasmine is edible, but the very similar false jasmine is poisonous. Garden pea flowers are edible, sweet pea flowers are not. Poppy seeds are food, but the poppy flower itself is toxic
The short version
- A cut flower is legally not food, so it has no residue limit and no pre-harvest interval. Treatment is allowed right up to harvest and after it
- Ninety Belgian bouquets carried 107 substances, about ten per bouquet, with roses averaging 14 substances and 26 mg/kg in total
- A second sample of 237 flowers from Austria and Germany: residues in 97%, harmful substances in 72%, persistent ones in 85%
- The residues transfer on contact. Florists’ gloves averaged 37 substances, and their urine averaged eight residues per sample
- After cutting comes a second wave: silver, bactericides, a fungicide dip, wax coatings
- Washing will not save it. The compounds are fat soluble, part of the load sits inside the tissue, and a petal cannot be scrubbed
What surprises me about this topic is not the amount of chemistry on flowers. It is how readily we move decor onto food as soon as it looks good. A bouquet is an industrial product engineered to arrive flawless, and that engineering is precisely why it carries what it carries.
