On 7 September, Michelin presented its second Grape selection, this time for Bordeaux. Its corporate press release counts 83 selected estates. Another Michelin article opens with 84 and announces 38 producers with one Grape, while the corresponding list contains 37 names. Add the categories together and the total is still 83; I counted the full list myself. On 12 September, five days after publication, the error was still online.
Anyone can make a mistake. Michelin, however, sets a higher standard for itself, precisely because the Grape needs to project accuracy and inspire trust. That makes the error relevant, although it is hardly the main issue. The more interesting question is how Michelin actually assesses those 83 selected producers.
Château Latour is nowhere to be seen
Absence is at least as fascinating as inclusion in the Michelin Grape because Michelin does not publish why a wine producer falls outside the selection. The 1855 classification of red Bordeaux today contains five Premiers Grands Crus Classés: Château Lafite Rothschild, Château Latour, Château Margaux, Château Haut-Brion and Château Mouton Rothschild. Michelin gives Lafite three Grapes. Margaux, Mouton Rothschild and Haut-Brion receive two.
Château Latour is absent altogether, including from the ‘Selected’ category.
So are Château Ducru-Beaucaillou, one of the Médoc’s most highly regarded deuxièmes crus, and Château Léoville Poyferré, another second growth in Saint-Julien. Château Pavie in Saint-Émilion is missing as well. One participant on WineBerserkers wrote: I thought the Burgundy one was pathetic. This one shows total ineptitude.
None of this means Michelin should necessarily have selected these producers. A new hierarchy has a reason to exist precisely when it challenges established reputations. A system that simply reproduces the 1855 classification adds little.
With names of this stature, however, you want to know why they are absent. Michelin does not say whether Latour was visited, assessed, fell outside the scope, declined to cooperate or failed to meet the criteria sufficiently. Consumers therefore cannot infer anything from the absence itself. I raised exactly that problem after the Burgundy selection: as long as Michelin does not explain why an estate is missing, the market will fill in the gap itself. With Château Latour, Bordeaux now provides an absentee of the highest possible calibre.
Montrose above Mouton is precisely what makes this interesting
The Grape disrupts existing wine hierarchies. That makes explanation and substantiation more important. Château Montrose and Château Léoville Las Cases receive three Grapes, while Mouton Rothschild, Margaux, and Haut-Brion receive ‘only’ two. A new classification should create room for current quality to outweigh historical status and inherited image.
Michelin stresses that the Grape is not a ranking of individual producers. Within a single category, that is true: there is no number one or number two. The categories themselves, however, clearly form a hierarchy. Three Grapes rank above two, two above one, and one above ‘Selected’. Michelin is therefore making a meaningful statement about quality level and consistency. What it does not explain is why Montrose receives three Grapes and Mouton two.
Drouhin finally answers an old question
After my previous article about the Grape, I finally received a response from Maison Joseph Drouhin. It belongs substantively to the Burgundy selection, but answers a question that remained open at the time. Domaine Arnoux-Lachaux, also in Burgundy, had previously said it had never received Michelin. Drouhin responds differently: the house does not deny that a visit took place, but cannot confirm one either.
That matters because Michelin has explained something important about its working method. Unlike restaurant inspectors, Michelin’s wine inspectors do not visit anonymously. They announce themselves and request access to the vineyard, cellar, and tasting room. Michelin told WineNews that its wine inspectors identify themselves and ask for access to the vineyard, cellar, and tasting room. The San Francisco Chronicle received the same confirmation from Michelin. That gives Arnoux-Lachaux’s answer additional weight: an announced visit is difficult to miss entirely.
With Drouhin, the opposite problem arises. Communications officer Poppy Image writes on behalf of the house: ‘We are sure they were hosted by our estate on several occasions without us knowing; they probably also purchased wine for their panel to taste.’ That sits awkwardly with Michelin’s own description of the process. If inspectors identify themselves and request access, a large house such as Drouhin should in principle be able to trace such a visit.
Drouhin is meanwhile actively helping to promote the Michelin Grape. The maison supplied two photographs for Michelin’s website and uses its distinction in its communications. Both parties therefore benefit from the same publicity: Michelin draws on the standing of a respected Burgundy house, while Drouhin benefits from Michelin’s authority and reach.
Two producers, the same uncertainty
Arnoux-Lachaux and Joseph Drouhin respond to Michelin in entirely different ways, yet arrive at the same point: neither producer can reconstruct the assessment. Arnoux-Lachaux says Michelin never visited. Drouhin assumes a visit probably took place somewhere among the many visitors it receives each year. One producer disputes that a visit happened; the other welcomes visitors yet cannot identify it either.
Drouhin also raises another question: which grapes and vineyards were actually assessed? The maison owns vineyards but also buys grapes under long-term contracts. Poppy Image writes: ‘All our grape sources are treated the same, be it on the vine or in the cellar.’ In her view, Michelin therefore has no reason to separate the two, ‘because we do not either’.
That is a clear answer about Drouhin’s own business practice, but it does not answer the question about Michelin’s approach. Did Michelin assess Drouhin’s estate vineyards separately? And the purchased grapes separately? Or was Maison Joseph Drouhin assessed as a single entity? The published information does not say.
Robert Parker complicates the question of visits
The role of Robert Parker Wine Advocate within the Grape is clearer today than it was at launch. In an interview with Corriere della Sera on 5 December 2025, international director Gwendal Poullennec explained that former Wine Advocate tasters are now Michelin employees. Some continue to assess individual bottles and award points for Robert Parker; others work as Michelin wine inspectors and assess producers as a whole. Some employees work for both publications, according to Poullennec, but in different roles.
Michelin also uses Wine Advocate’s historical database of more than 500,000 wine reviews to assess producers’ performance over time. In the small print at the bottom of the Michelin Guide website, it also states that personal data may be shared with other ‘Michelin affiliates’, including TabletHotels and Robert Parker.
The connection between Michelin and Robert Parker is therefore clear. Less clear is how previous visits, tasting notes and historical assessments feed into the Grape decision. Does an earlier visit by a Parker reviewer count as part of Michelin’s assessment? Is an old tasting note reassessed under the continuity criterion? And how does that fit with Michelin’s own statement that wine inspectors identify themselves to producers?
What Michelin has published about its working method does not fully answer those questions. It therefore remains unclear which visits to producers such as Arnoux-Lachaux and Drouhin actually formed part of the Michelin assessment. A producer may already have been reviewed for years by people who now work for Michelin. Earlier Wine Advocate reviews are also used to assess performance over time.
Then look at the assessments themselves
If the process behind the scenes is difficult to reconstruct, perhaps the published producer profiles can show how Michelin applies its five criteria. I therefore compared four Bordeaux producers with three Grapes: Château d’Yquem, Château La Conseillante, Petrus and Château Cheval Blanc.
The profiles treat the same five criteria with markedly different emphasis. Sometimes a criterion receives concrete substantiation, sometimes only a few general sentences, and sometimes it almost disappears from view.
Yquem: mostly the wine in the glass
At Château d’Yquem, one criterion dominates: balance. Michelin describes at length how sweetness and acidity relate to each other, then adds saffron, candied citrus and dried stone fruit. At Yquem, the substantiation shifts strongly towards a classic tasting note. Agronomy receives only cursory treatment; identity is mainly supported through history and generations of winegrowers.
La Conseillante: technique and an upward trajectory
Château La Conseillante also receives three Grapes, yet its profile looks completely different. Concrete technical choices dominate here: harvest in 15-kilogram crates, a night in cold storage, manual pigeage and a declining proportion of new oak.
The criterion of continuity is particularly striking. Michelin writes: ‘Recent vintages confirm the estate’s upward trajectory, with 2023 emerging among the most compelling wines of the vintage.’ An upward trajectory is something different from proven continuity over a longer period.
Petrus: continuity suddenly becomes tangible
Petrus offers a much more concrete substantiation of continuity. The transition from Jean-Claude to Olivier Berrouet is described by name, and Michelin links that generational change to technical choices, including the use of new oak, the choice of coopers and the duration of ageing.
Agronomy, by contrast, receives hardly any attention. The emphasis shifts again, although the distinction remains the same.
Cheval Blanc: agronomy comes first
Château Cheval Blanc also receives three Grapes, with yet another kind of profile. Agronomy and technique receive considerable attention. Michelin describes hedgerows, fruit trees, massal selection and measures aimed at soil life. Balance receives far less space, while continuity is barely substantiated separately.
The four profiles use the same five criteria but emphasise different elements. Balance dominates at Yquem, technique and recent quality improvement at La Conseillante, continuity at Petrus and agronomy at Cheval Blanc. A common yardstick therefore remains difficult to identify.
Five criteria still do not make a verifiable method
Bordeaux confirms a problem that, in Burgundy, was mainly visible through the lack of methodological explanation. The five criteria are known, but their relative weight is not. Nor is the minimum evidence required for each criterion clear. The producer profiles do not reveal a consistent application either.
That does not mean Michelin assesses producers arbitrarily. There is no evidence for that either. It does mean that an assessment cannot be reconstructed. Michelin may have extensive tasting notes, visit reports, vineyard data and assessments covering several vintages. As long as that information remains out of sight, consumers, and producers have to trust the result without being able to follow how Michelin reached it.
What does the consumer gain from it?
For consumers, the Michelin Grape has an obvious appeal. The wine world is complicated; classifications accumulate and reputations can be decades old. One, two or three Grapes make that landscape easier to navigate. The symbol brings together a large amount of technical and historical information in a form that can be used quickly in a shop, restaurant or on a wine list.
That is also Michelin’s greatest strength. The name is already familiar and has built authority in restaurants and hotels. The Grape therefore starts with an advantage that a new rating system would normally have to earn first. Its first selections are immediately visible and relevant.
Trust grows when consumers find that Michelin consistently applies the same quality standard and that the outcomes make sense alongside one another. Consumers are unlikely to care much about the mechanics themselves; the reliability of the result matters more. And the more regions Michelin adds, the easier it will become to see whether three Grapes in Bordeaux, Burgundy and later other regions genuinely represent the same level.
Trust still has to be earned
The first two selections do not yet provide a complete answer. The differences in explanation, visits, and substantiation remain too large. At the same time, the system is young. Michelin therefore still has room to make its assessment practice more consistent and explain more clearly how the five criteria are applied in practice.
That is the authority Michelin can still build. The name opens the door. The next selections will determine whether the Grape can make that authority work in wine as well.
Sources
MICHELIN Guide, article announcing the Bordeaux launch (7 September 2026, accessed 12 September 2026); Michelin Group, corporate press release (8 September 2026); MICHELIN Guide, official Bordeaux selection; MICHELIN Guide, Wine Selections overview and privacy notice; written responses from Maison Joseph Drouhin to Winecastr (September 2026); Corriere della Sera, interview with Gwendal Poullennec (5 December 2025); WineNews, interview on the working method of the wine inspectors; San Francisco Chronicle, reporting on non-anonymous wine-estate visits; own count of the full Michelin list; earlier Winecastr articles: ‘The Michelin Grape gives wine buyers trust. But based on what?’ and ‘Michelin bepaalt nieuwe pikorde in Bourgogne’ (in Dutch).
This article first appeared in Dutch on Winecastr.com. The English version was produced with AI assistance and edited by the author.