What is a good Mediterranean diet score? The 14 checks behind the 0–14 number
If you have taken a Mediterranean diet questionnaire and come out with a number, the obvious next question is what that number is supposed to mean. The short version: in the published literature, a MEDAS score of 10 or more of the 14 criteria is the boundary most analyses use for “high adherence”, 6 to 9 of the 14 is usually grouped as moderate, and 5 or fewer as lower. Those are conventions for sorting study participants into groups, not a pass mark — and the more informative answer is not the total at all, but which of the fourteen checks you met.
Here is the whole instrument, mechanically.
The score is fourteen yes/no checks, not a grade
MEDAS — the Mediterranean Diet Adherence Screener — was developed for the PREDIMED trial and published by Schröder et al. (J Nutr, 2011). It asks fourteen questions about habitual intake, and each one is a threshold: you either clear it or you don’t. Clearing it is worth exactly one point, and no item is worth more than any other. That flat structure is why the whole thing takes about two minutes and needs no food diary — and it is also why a bare total tells you less than the item-by-item breakdown does.
| # | Criterion | Scores a point when |
|---|---|---|
| 1 | Olive oil used as the main culinary fat | Yes |
| 2 | Olive oil, total per day (cooking, salads, everything) | ≥ 4 tablespoons |
| 3 | Vegetables | ≥ 2 servings per day |
| 4 | Fruit | ≥ 3 servings per day |
| 5 | Red or processed meat (beef, burgers, sausages, ham) | < 1 serving per day |
| 6 | Butter, cream, or margarine | < 1 serving per day |
| 7 | Sugar-sweetened or carbonated drinks | < 1 per day |
| 8 | Wine | ≥ 7 glasses per week |
| 9 | Legumes (beans, lentils, chickpeas, peas) | ≥ 3 servings per week |
| 10 | Fish or seafood | ≥ 3 servings per week |
| 11 | Commercial, non-homemade sweets or pastries | < 3 per week |
| 12 | Nuts, including peanuts | ≥ 3 servings per week |
| 13 | White meat (chicken, turkey) usually preferred over red | Yes |
| 14 | Sofrito — tomato, onion and garlic cooked in olive oil | ≥ 2 times per week |
Two structural details worth knowing before you read your own number:
Four of the fourteen criteria are “less is better” — items 5, 6, 7 and 11 score when intake is below a threshold. A blank form therefore scores 4 of the 14 rather than 0, because zero red meat, zero butter, zero sugary drinks and zero pastries all clear their thresholds. A partial answer set is not a low score; it is a meaningless one. (This is why our own app refuses to compute a total until all fourteen are answered.)
Item 8 is a criterion of the published instrument, not a suggestion. MEDAS was built in a Spanish trial population where wine with meals was customary, and the screener records that as one of its fourteen thresholds. We report the criterion because leaving it out would misdescribe the instrument. Describing a threshold is not encouragement to meet it: nothing here, and nothing in the app, suggests taking up or increasing drinking.
Where the “10 or more” boundary comes from
It comes from how the trial’s data was analysed, not from a clinical cut-off. PREDIMED analyses commonly split participants into adherence groups to compare them, with 10 of the 14 criteria as the lower edge of the high-adherence group. It is a line drawn to make a comparison legible. Nothing mechanical happens when a score moves from 9 to 10 — the difference is one criterion, and which criterion it was matters more than the total shifting by a point.
Our app uses those same conventional boundaries and labels them plainly: 10 or more of the 14 criteria as high adherence, 6 to 9 as moderate, 5 or fewer as lower.
Why two people with the same score can eat quite differently
Because the items are unweighted, the same total can be assembled in very different ways. Someone who clears the vegetable, fruit, legume, fish and nut criteria but cooks with butter and drinks a soft drink daily can land on the same total as someone who clears the olive-oil, sofrito, white-meat and low-sugar criteria while eating little produce. The screener will report the same number for both.
That is the honest limit of a single figure, and it is the argument for reading the breakdown. Nine of the 14 criteria met tells you how many; only the item list tells you which five you missed, and whether you miss the same ones every time.
What the number does not tell you
- It is self-reported. “Servings per day” is a recall estimate. The instrument was designed to be fast and rough, not precise, and its own validation study measured it against fuller dietary assessment rather than treating it as ground truth.
- Every criterion is a step. Two servings of vegetables scores; 1.9 does not. Ordinary week-to-week variation can move a total by a point or two with nothing real having changed underneath it.
- It was developed and validated in a specific population — Spanish adults at elevated cardiovascular risk (Schröder et al., 2011). Serving conventions, and foods like sofrito, reflect that setting rather than a universal one.
- It describes a dietary pattern. The output is a count of criteria met. It is not an assessment of a person, and no score here is a diagnosis or a prediction about anyone.
On outcomes, the honest statement is a statement about the trial: in PREDIMED, participants assigned to the Mediterranean-diet interventions had roughly 30% fewer major cardiovascular events than the control group over the follow-up period (Estruch et al., NEJM, 2013; 2018 re-analysis). That is a published finding about that trial’s groups, in that trial’s population, over that trial’s follow-up — not a projection that any individual score generates.
Computing it
We built Mediterranean Diet Score to run exactly the fourteen checks above: it asks the questions in plain English, applies the thresholds in the table, and shows which criteria you met and which you missed rather than only the total. It is free, there is no account, and re-checks are kept in your own browser’s storage so you can see whether the same items keep coming up.
Sources: Schröder H et al., “A short screener is valid for assessing Mediterranean diet adherence among older Spanish men and women,” J Nutr, 2011. Martínez-González MA et al., PLOS One, 2012 (10.1371/journal.pone.0043134), which publishes the full 14-item instrument under a Creative Commons Attribution licence. Estruch R et al., “Primary prevention of cardiovascular disease with a Mediterranean diet,” NEJM, 2013; 2018 re-analysis. This post describes published science and software output — not medical or dietary advice.
