Most advice about training your nose is invented. Here is what the evidence actually supports, what it does not, and a programme you can run on your own.
Smell training works, and we have to be careful about what that means
The clinical protocol is called olfactory training and it was formally described by Thomas Hummel and colleagues at Dresden in 2009.
The method is unglamorous. Four strong odours, smelled deliberately twice a day for twelve weeks. The originals were phenyl ethyl alcohol for rose, eucalyptol for eucalyptus, citronellal for lemon and eugenol for cloves.
Reported effectiveness runs from 11% to 68% across studies after twelve to sixteen weeks, and imaging work has found changes in functional connectivity afterwards, which suggests the olfactory system reorganises with use.
Now the caveat, and it matters. That research studied people with olfactory dysfunction, meaning damaged or diminished smell, often after illness. It is evidence that a compromised sense can be rehabilitated. It is not evidence that a normal nose can be trained into an expert one, and there is not good clinical work on that specific question yet.
What follows is therefore part evidence and part craft tradition, and we have marked which is which.
What actually improves with practice
Not sensitivity. Almost certainly not sensitivity.
What improves is naming and recall. The trained difference between a perfumer and you is not that they detect more molecules. It is that they have an organised mental library, so when something arrives they can say "that is orris, and there is too much of it" instead of "that smells sort of powdery."
The naming problem is real, and it is not your fault
There is a documented phenomenon here. Most people find it profoundly difficult to name familiar smells, even when their perception is unimpaired and they can name objects by sight without any trouble. The failure sits in the link between the olfactory system and language, not in the nose.
You know this feeling. You smell something, you are certain you know it, and the word will not come. That is sometimes called the tip-of-the-nose state, and it is the ordinary condition of an untrained smeller.
But it is not a fixed human limit, and this is the encouraging part. The claim that humans are simply bad at naming odours has been challenged, because there are languages with large dedicated smell vocabularies where naming odours is straightforward. Jahai speakers on the Malay Peninsula name odours far more readily than English speakers, because Jahai has abstract words for smell qualities in the way English has abstract words for colour.
So the deficit is linguistic and cultural, not anatomical. English gives you almost nothing: we describe smells by naming their sources ("lemony," "smoky") or by borrowing from taste ("sweet"). A perfumer's training is, more than anything, the construction of the vocabulary English never gave them.
Training here means building a lexicon and attaching it to things you have smelled, not training the nose itself. Every exercise below asks you to say or write the word.
There is also evidence the system is plastic in a broader way: olfactory memory training has been studied for transfer to untrained visual memory tasks, which suggests smell sits closer to general memory machinery than its reputation implies.
Our guide to the perfumers makes the same point from the other direction. Several of the perfumers there came in through chemistry rather than through some inherited gift, and Jean Carles composed two classics after going anosmic, working from memory of what materials do. The library is the skill. The nose is the input device.
Your nose can be measured, which most people do not know
Olfactory ability is a measured quantity with published normal ranges, not a mystery.
Two instruments dominate. Sniffin' Sticks presents odours on reusable pen-like dispensers and tests three separate things: your threshold (how little of a substance you can detect), your discrimination (can you tell two apart), and your identification (can you name it from a list). The three add up to a TDI score. The University of Pennsylvania Smell Identification Test, or UPSIT, is 40 scratch-and-sniff items with four choices each, and it tests identification only.
The normative study is large. Hummel and colleagues published TDI norms from more than 3,000 subjects. The tenth-percentile TDI score was 30.3 for ages 16 to 35, 27.3 for 36 to 55, and 19.6 for over 55. Women outperformed men on all three components. And the component that falls off fastest with age is the threshold, not identification.
That last detail is the encouraging one for this guide. Your sensitivity is largely given to you. Your ability to identify and discriminate holds up far better, and those are the two the exercises below actually work on.
The decline is real, and the numbers are more useful than a vague sense of dread. The US National Institute on Deafness and Other Communication Disorders puts the prevalence of smell impairment at 4% of Americans aged 40 to 49, 11% at 50 to 59, 13% at 60 to 69, 25% at 70 to 79, and 39% at 80 and over (US data; we found no equivalent Canadian series).
So most people reading this have an ordinary nose and will keep one for decades. If you are over seventy and finding fragrance harder to read, you are in a large group and it is not your imagination. If you are thirty and think you have a bad nose, you almost certainly have an untrained vocabulary instead. Those are different problems and only one of them has a fix here.
The programme
Twelve weeks, because that is the interval the clinical work uses. Ten minutes a day.
Weeks 1 to 4: single materials, named
Get raw materials. Perfumery supply houses sell small quantities of the standard ones, and a starter set of ten to fifteen is enough to change how you smell everything.
Start with the ones our materials guide covers, because the reading gives you somewhere to file them: bergamot, lavender, rose, jasmine, orris, vetiver, patchouli, oakmoss, tonka or coumarin, vanilla, cedar, sandalwood.
Smell one. Say its name out loud. Say one thing it reminds you of that is not a fragrance: a place, a food, a room. That second step is the one people skip and it is the one that builds recall.
Expect disappointment. Most raw materials smell worse on their own than you imagine. Oakmoss is bitter and inky. Civet-adjacent musks smell wrong. Even jasmine absolute has something faintly rotten in it: that is the indole. This is normal and it is the first real lesson: perfumery is not made of nice smells arranged nicely.
Weeks 5 to 8: pairs and contrasts
This is the Jean Carles method, simplified. Carles taught raw materials in ordered contrasting pairs, tabulated by class, so students built a map rather than a heap. Smell two things together and name the difference:
- Bergamot against lemon. Which is rounder?
- Cedar against sandalwood. Which is dry, which is creamy?
- Rose against geranium: geranium is the cheap stand-in for rose and shares much of its chemistry.
- Vetiver against patchouli. Both earthy. Say how they differ.
The naming of the difference is the exercise. Doing it aloud or in writing works considerably better than doing it in your head.
Weeks 9 to 12: structures
Now put fragrances rather than raw materials side by side. Each pair below is a clean example of one idea, and where we carry a real 2ml decant of both fragrances in a pair, we have linked it, so you can order the actual pair and smell it yourself rather than take our word for it.
- A trio of overdoses. Bandit, Vent Vert and Fracas are all Germaine Cellier compositions, each built by taking one material and overdosing it far past what anyone else was doing at the time. We do not carry any of the three as decants, but the lesson works from description alone if you can find samples elsewhere: one material overdosed, three times, three different results. Learn what an overdose feels like.
- A fougère pair, six years apart. Guy Laroche Drakkar Noir (1982) is a straightforward aromatic fougère. Davidoff Cool Water (1988) sits on the fougère/aquatic boundary our fragrance families guide describes, aimed at a completely different room. Why this pair: it shows how much room one structure leaves for two perfumers to disagree.
- A vetiver pair, forty-seven years apart. Guerlain Vétiver (1959) and Terre d'Hermès (2006) both draw on the same root. Why this pair: Guerlain's version puts vetiver at the centre with citrus and tobacco around it; Terre d'Hermès uses it as one base note among several, in service of a mineral, citrus-forward composition rather than a vetiver composition. Same root, opposite jobs.
- Two fragrances with one perfumer in common, nine years apart. Angel (1992, Olivier Cresp with Yves de Chiris) and Dolce & Gabbana Light Blue (2001, Cresp alone) could hardly smell less alike: one is a dense gourmand built on a patchouli-heavy, chypre-adjacent base, the other a citrus-and-jasmine sheer. We do not carry Angel as a decant, so this one works best as a description exercise unless you can find it elsewhere. Why this pair: it shows what a perfumer's signature actually is when the two fragrances share almost no notes.
Smell each pair back to back, and write down what changed. Not whether you liked it.
Keeping notes, and what to write
A scent journal sounds precious and it is the single highest-return habit here. Rather than a printed template, copy this structure into any notebook, and record, every time:
- What it is, and the batch code if you have it.
- Where and when, including the weather, because heat changes everything.
- What you smell in the first minute, at thirty minutes, and at four hours. Three entries, not one. Our notes and accords guide covers why the first minute lies.
- What it reminds you of. A place, a person, a material. This is the recall hook.
- A verdict, in one blunt line. "Too sweet." "Would wear to work." "Good, not for me."
Do not write a review. You are building an index, not publishing.
After three months you will be able to look back and find patterns you did not know you had: a family you consistently reach for, a material you consistently dislike. That matters more than any recommendation anyone can give you.
Three rules that will save you
Smell less, more often. Ten minutes daily beats two hours on Saturday, and our guide to testing perfume explains the fatigue that makes long sessions worthless.
Do it in different conditions. Same fragrance in January and in July is two different fragrances.
Smell things that are not perfume: cut a lemon, open the coffee, crush a leaf from the garden. Perfumery is built out of references to the real world, and if your library only contains perfume then every fragrance can only ever smell like other fragrances.
A home version of the discrimination test
Borrowed from the structure of the clinical one: run it at week 1 and again at week 12 so you have a before and an after.
Get someone else to prepare it. Three blotters, two of them sprayed with the same fragrance and one with something different. You smell all three and name the odd one out. Twelve rounds, written down, no talking during. Score out of twelve.
Start with easy pairs from opposite families and work toward hard ones: two vetivers, two ambers, a fragrance against its own flanker. The point is not the score in isolation. It is the same score twelve weeks later, on the same pairs, prepared by the same person.
If it has not moved, the training is not working and you should change it rather than keep going.
Part of Nose to Bottle, our complete fragrance manual. Read it in order here.