Balancing a vegetarian plate: what the official Swiss model actually says

In Switzerland the official reference is not quarter-quarter-half. Since 2025 the Federal Food Safety and Veterinary Office (FSVO) has published the “balanced plate” in proportions of two fifths fruit and vegetables, two fifths starchy foods and one fifth protein-rich foods. The official protein share is therefore smaller than the quarter you read everywhere, and the starch share is larger.

That is the first thing to fix, because the version repeated everywhere claims a quarter protein, a quarter starch and half vegetables, and credits it to the Harvard model. Neither model says exactly that. What follows is the actual set of proportions, converted into weights you can put on a scale, and more importantly the things these models do not settle at all in a vegetarian diet.

The three models disagree in ways that matter

Several reference plates exist and they diverge on points that change how you cook.

The FSVO balanced plate, updated in 2025 with the Swiss Society for Nutrition and Health Promotion Switzerland, uses two fifths, two fifths, one fifth. Fruit and vegetables share one section, starchy foods take an equal section, and protein-rich foods take a section half that size. The official FAQ adds that these proportions “are not to be taken literally” and give an order of magnitude valid for most adults.

The Healthy Eating Plate from the Harvard T.H. Chan School of Public Health puts half the plate in vegetables and fruit, a quarter in whole grains and a quarter in protein foods. The fruit detail matters: read “half vegetables” and you have quietly dropped half the contents of that section. Harvard also adds a point that most guides skip, namely that potatoes and chips do not count as vegetables in this model because of their effect on blood sugar. They sit with the starchy foods.

The NHS does not use a plate at all. The Eatwell Guide describes proportions across the whole diet: just over a third fruit and vegetables, just over a third starchy carbohydrates. It states plainly that you do not need to achieve this balance at every meal, but over a day or even a week.

All three are teaching tools rather than prescriptions. They agree on one broad idea, that roughly half of what you eat should be non-starchy plant food, and they disagree on the detail. If you cook in Switzerland, the two fifths, two fifths, one fifth marker is the one that matches the guidance in force here.

What does a fifth weigh? Turning fractions into grams

You cannot cook a visual fraction. Nobody tips “two fifths of rice” into a pan. The Swiss portion sizes, on the other hand, come with numbers, and those numbers are what you actually work with.

For fruit and vegetables, one portion is 120 g, roughly a handful. The daily recommendation is 5 portions split into 3 of vegetables and 2 of fruit, which comes to 600 g a day. The WHO sets a similar floor: at least 400 g of fruit and vegetables a day for everyone over 10, 250 g for children aged 2 to 5 and 350 g for children aged 6 to 9.

For grain products and potatoes, the recommendation is 3 portions a day, at least half of them wholegrain. One portion is 75 to 125 g of bread or dough, or 200 to 300 g of potatoes, or 45 to 75 g of flakes, pasta, rice, crispbread, polenta, couscous, buckwheat, quinoa or flour, measured dry. That last figure is the most useful one in a kitchen: 45 to 75 g of dry rice per person is a good deal less than what most people cook.

For protein-rich foods, the recommendation is one portion a day, and a portion is 60 g of dry legumes, or 120 g of tofu, tempeh or plain seitan, or 30 to 40 g of soya granules, or 2 to 3 eggs. On top of that come 2 to 3 portions of dairy a day, where one portion is 2 dl of milk, 150 to 200 g of yoghurt, quark or cottage cheese, 30 g of semi-hard or hard cheese, or 60 g of soft cheese.

Two groups round out the plate without appearing on it. Nuts and seeds: one small unsalted handful a day, 15 to 30 g. Oils: 2 tablespoons of vegetable oil a day, about 20 g, of which at least one should be rapeseed oil, plus a maximum of 10 g of butter or margarine. And a drink: 1 to 2 litres of unsweetened drinks a day, with at least one glass or cup of 2 dl at each main meal.

The FSVO vegetarian factsheet translates all this into per-meal terms: at least 1 glass or cup of unsweetened drink, 1 to 2 portions of fruit and vegetables, 1 portion of grain products or potatoes, and 1 portion of dairy, legumes, eggs or another protein source.

One vegetarian plate, built from numbers

Take an evening meal for one person, assembled from those figures rather than from fractions.

Weigh out 60 g of dry green lentils, which will yield roughly 150 g cooked. Add 60 g of dry brown rice, staying at the lower end of the 45 to 75 g range because the lentils bring starch too. Allow 240 g of vegetables, which is two portions: say 120 g of courgette and 120 g of carrot, roasted. Finish with a tablespoon of rapeseed oil, 15 g of pumpkin seeds and a glass of water.

Cooked, that lands at roughly 500 g on the plate. Visually the vegetables and lentils dominate and the rice takes up far less room than you would expect. That is normal, because a visual fraction and a dry weight tell different stories: rice and pasta roughly triple in weight as they cook, while roasted vegetables lose weight.

What the plate fraction does not tell you about protein

Here is the most useful finding in this article, and one that plate models almost never spell out.

A study published in Frontiers in Nutrition in 2026 calculated the DIAAS of every meal recorded in 4-day food diaries kept by 193 New Zealand vegans. DIAAS scores protein quality by accounting for digestibility and indispensable amino acids. Eight meal clusters emerged. The higher-scoring meals shared a clear signature: a legume-to-grain ratio of roughly 2 to 1 by weight. The lower-scoring meals contained more fruit, grain foods, potatoes and sugary condiments, with a significantly smaller share of legumes (p < 0.001).

The chemistry behind it is simple. Grains, nuts and seeds supply mostly sulphur amino acids. Legumes supply mostly lysine. Neither family is sufficient alone, and the ratio between them matters more than the space each occupies on the plate.

The most uncomfortable result sits elsewhere in the paper: low protein quality meals also turned up in people whose total daily protein intake was high. A day that looks fine on paper can hide unbalanced individual meals. That is precisely what a plate model is meant to fix, and precisely what a visual fraction does not measure.

In the meal described above, 60 g of lentils to 60 g of rice gives a 1 to 1 ratio. To move closer to the profile seen in the study, push the lentils to 80 g and drop the rice to 45 g. That is a trivial change in the kitchen, and it concerns the ratio between two ingredients rather than the area they cover.

Should you worry about protein quality scores at all?

No, and presenting DIAAS as a verdict would be dishonest. Two 2026 papers put serious limits on how far it should be pushed.

A critical review in Comprehensive Reviews in Food Science and Food Safety shows that PDCAAS and DIAAS both evaluate foods in isolation, identifying a single limiting amino acid per source. This approach systematically underestimates protein adequacy in real diets, where several sources complement each other. Applying a framework that pools amino acid contributions across the whole day, the authors obtain corrected values 18 to 30 per cent higher than food-by-food summation.

A review in Current Nutrition Reports goes further: scores based on amino acid composition correlate poorly with health outcomes. Animal proteins score higher, but those differences do not consistently translate into better cardiometabolic or mortality outcomes. In Western populations, where protein intake generally exceeds physiological requirements, modest differences in digestibility have limited clinical relevance.

The practical conclusion fits in one sentence: put legumes in your meals regularly, roughly twice as much as grains where you can, and stop worrying about the rest. Swiss guidance asks for legumes at least once a week, which is a very low floor for a vegetarian. Our guide on combining grains and legumes effectively covers the pairings that work in practice.

As for the 0.8 g per kilo and 1.2 to 1.5 g per kilo figures that circulate widely: they are plausible and broadly consistent with the literature, but they carry no source and they apply to very different situations. The complete plant protein guide handles requirements together with the matching portions.

The order you eat in changes the glycaemic response

A plate model describes what sits on the plate and never the order you eat it in. That is a shame, because the order has a measurable effect.

A randomised crossover trial published in Diabetes, Metabolic Syndrome and Obesity in 2024 gave 18 healthy adults two identical isocaloric meals in two different orders: a standard mixed meal on one occasion, vegetables and protein first followed by carbohydrates on the other. The incremental area under the glucose curve over 120 minutes was 40.9 per cent lower (p = 0.03) with the second sequence, and the insulin area 31.7 per cent lower (p = 0.023). Fullness was rated higher at 60 and 120 minutes.

A systematic review in Clinical Nutrition Research in 2026 screened 2,442 records and retained six studies covering 107 participants aged 20 to 36.7. Most reported the same effect: eating vegetables, fruit or protein-rich foods before carbohydrate-rich foods reduces the postprandial glucose response.

Keep the finding in proportion. The samples are small, the participants young and healthy, and the review authors themselves call for larger and longer trials. These are acute effects measured over two hours, not a demonstration of long-term benefit. But it costs nothing: starting with the salad or the roast vegetables before the rice requires no recipe change whatsoever. If the topic interests you, our article on the glycaemic index from a plant-based angle takes it further.

One-pot meals, the blind spot in every plate model

Most vegetarian meals look nothing like a compartmented plate. They are curries, risottos, gratins, thick soups, composed salads, bowls, pasta. The visual model stops being readable.

The FSVO FAQ addresses this case directly and its method is the right one: for a single dish such as a pizza, lasagne, vegetable tart or pasta salad, look at the quantity of each food group used in the recipe itself, then top the meal up if needed with a salad, a slice of bread or a portion of cheese.

Which means balancing at the shopping and prep stage rather than at the serving stage. For a chickpea curry you do not try to draw sections in the pot, you check what went into it: how many grams of chickpeas, how much dry rice, how many vegetables. A dish for four built on 240 g of dry chickpeas, 200 g of dry rice and 800 g of vegetables meets the markers without a single visible fraction.

One imbalance shows up more often than any other in vegetarian one-pot cooking: the starch share swells and the vegetable share evaporates. Pasta with pesto, risotto, polenta and thick-crust tarts are dishes where vegetables tend to become decorative. The FSVO factsheet says this outright, that vegetables should not be treated as mere decoration. Our article on buddha bowls shows how to build a single dish in layers while keeping the proportions visible.

What no plate fixes: four nutrients to watch

Getting the food groups right does not guarantee micronutrient coverage. This is the fundamental limit of every plate model, and it bites harder on a vegetarian diet.

Iron. The NHS puts the requirement at 8.7 mg a day for men aged 19 and over, 14.8 mg for women aged 19 to 49, and 8.7 mg for women aged 50 and over. The gap between the first two figures is large, and it explains why iron is a completely different question depending on who is asking. Non-haem iron from plants absorbs less readily than haem iron, which makes the optimisation techniques worth knowing: our guide on plant iron absorption covers which ones actually work.

Calcium. 700 mg a day for adults aged 19 to 64, per the NHS. One detail there deserves attention because it travels badly: the NHS lists green leafy vegetables such as curly kale, but explicitly excludes spinach, which contains plenty of calcium that the body cannot fully absorb. The sources for covering calcium without dairy are narrower than they look.

Iodine. 140 micrograms a day for adults. This is where vegetarianism creates a genuine problem, because the sources the NHS lists are cows’ milk, dairy products, eggs, sea fish and shellfish. Plants contain iodine, but the level in cereals and grains depends directly on the iodine in the soil they grew in. On a strict vegan diet the NHS suggests considering fortified foods or a supplement. It is the nutrient that articles on vegetarian balance handle worst. Our article on balancing iodine with seaweed and iodised salt develops the subject.

Vitamin B12. 1.5 micrograms a day for adults aged 19 to 64, per the NHS, which notes that B12 is not found naturally in fruit, vegetables or grains. An ovo-lacto vegetarian eating eggs and dairy regularly will get it. A vegan must supplement, without exception. The NHS also states that taking 2 mg a day or less in supplements is unlikely to cause harm. See our article on B12 requirements and reliable sources.

Vitamin D belongs on the list too, and it barely depends on the plate at all. The NHS recommends 10 micrograms a day from the age of 1, including during pregnancy and breastfeeding, and notes that from October to early March skin synthesis is no longer sufficient at those latitudes. Switzerland sits at comparable latitudes. No arrangement of the plate compensates for winter. Our article on keeping vitamin D up when the sun is missing deals with this.

Fruit, routinely overlooked

It should be said plainly: fruit goes unmentioned in many presentations of the plate model. Yet fruit shares the same section as vegetables in the Harvard model, and accounts for 2 of the 5 daily portions in Switzerland.

The FSVO factsheet is flexible about it: the vegetable portion can be supplemented or replaced by fruit. So you can build fruit into the meal, in a salad or as a side, or eat it separately. An apple after the meal counts as much as an apple in it.

One WHO point belongs here because it is poorly known: fruit juices, including those without added sugar, contain significant amounts of free sugars and intake should be limited. The NHS caps juice and smoothies at 150 ml a day combined. A glass of orange juice is not the equivalent of an orange.

On fibre, the WHO sets a target of at least 25 g a day of naturally occurring dietary fibre for everyone over 10, 15 g for ages 2 to 5 and 21 g for ages 6 to 9. A vegetarian diet built on legumes, whole grains and 600 g of fruit and vegetables clears that without special effort.

What the evidence supports, and what it does not

Plate guides routinely claim that half a plate of vegetables helps with weight management and that cruciferous vegetables have a “positive impact on health”, with no further detail. Here is the state of the evidence on two specific points.

On legumes, a meta-analysis in Nutrition Reviews in 2025 pooled 31 randomised controlled trials of non-soy legumes. Consumption lowered total cholesterol by 5.159 mg/dL (95% CI: -7.192 to -3.126) and LDL by 3.475 mg/dL (95% CI: -5.321 to -1.530). The effect on triglycerides was not significant overall, and no effect appeared on HDL. These are randomised trials, so the evidence level is solid, but the magnitude stays modest: a few milligrams per decilitre, not a transformed lipid profile.

On fruit and vegetables, a meta-analysis in Nutrients in 2026 pooled 20 observational studies covering 24,660 cases of prediabetes among 182,387 participants. Higher intake was associated with lower prediabetes risk (OR 0.82, 95% CI: 0.74-0.91). The detail is worth reading, though: the association was significant across the 16 cross-sectional studies and absent in the 3 cohort studies (OR 0.97, 95% CI: 0.92-1.02), with very high heterogeneity (I² = 91.9 per cent). Cross-sectional studies capture a snapshot and cannot establish direction of causality. The cohorts, which follow people over time, do not find the effect. The authors themselves conclude that more prospective work is needed.

Eating vegetables and legumes remains a good idea. But promising weight loss or diabetes prevention on that basis would go well past what the evidence supports.

Adapting the model, knowing it does not cover you

This point is critical, and it is missing from most presentations of the model, which cheerfully apply it to children, athletes and older adults.

The FSVO FAQ states in black and white that Swiss nutrition recommendations address healthy people aged 18 to 65, and that children, adolescents, pregnant women, people with medical conditions and people following particular diets, “vegetarian, vegan or athletes, for example”, are not covered. The FAQ adds that such people need to take specific considerations into account.

Which is to say: the typical reader of this article sits explicitly outside the scope of the model they are consulting. That does not invalidate the plate as a visual marker, but it does rule out presenting it as a complete answer.

Endurance athletes. Energy and carbohydrate needs rise, and increasing the starch share is reasonable. Under heavy training the total quantity matters more than the fraction. Our article on vegetarian diets and endurance sport covers this case.

Older adults. A narrative review in Frontiers in Nutrition in 2026 on sarcopenia argues that general protein recommendations may be insufficient past a certain age, and that higher intakes, spread better across the day, are needed to stimulate muscle protein synthesis. The authors note that animal proteins offer greater bioavailability, while concluding that strategic combinations of plant proteins can also meet amino acid requirements. They also stress resistance training, which is not a plate question. This is a narrative review rather than a meta-analysis, so the evidence level is intermediate.

Children. The WHO lowers the fruit and vegetable targets to 250 g for ages 2 to 5 and 350 g for ages 6 to 9, and fibre to 15 g and 21 g respectively. The concern about excess fibre in young children is real and points the same way. But since Swiss recommendations do not cover this age group, paediatric advice beats an adult plate model.

Weight loss. The plate is not a calorie restriction tool and was not designed as one. Increasing the vegetable share increases volume at low energy density, which is coherent, but none of the official sources consulted presents the model as a weight loss device.

Six mistakes that come up constantly

Confusing dry and cooked weight. 60 g of dry rice yields around 180 g cooked. Plenty of people weigh out 150 g of dry rice thinking it is one portion and end up with triple. It is probably the most common mistake and the one with the biggest effect on the real balance of the meal.

Counting potatoes as a vegetable. They belong with the starchy foods, and Harvard explicitly excludes them from the vegetable section. A plate of potatoes and green beans is not a mostly-vegetable plate.

Forgetting fruit. Two of the five daily portions. Not an optional dessert.

Treating juice as fruit. 150 ml a day maximum for juice and smoothies per the NHS, and the WHO counts their sugars as free sugars even with nothing added.

Believing every meal must be perfect. The NHS says the opposite: balance plays out over a day or a week. One unbalanced meal is not a problem. An unbalanced week is.

Treating tofu and lentils as interchangeable gram for gram. A portion is 60 g of dry legumes or 120 g of tofu. The ratio is one to two, because tofu carries far more water.

Breakfast, a special case

The FSVO factsheet is clear on a point most readers miss: the model applies to breakfast as much as to lunch and dinner. Ideally each of the three main meals contains a drink, fruit or vegetables, a starchy food and a protein-rich food.

In practice the common vegetarian breakfast in Switzerland is bread, jam and coffee. It delivers the starch and nothing else. A muesli with yoghurt or a fortified plant drink, some fruit and a handful of seeds ticks all four boxes with no effort. The FSVO’s explicit advice to round meals out with nuts and seeds finds its simplest application here. Our article on the balanced vegetarian breakfast develops these combinations.

Limits of the available data

Three caveats belong on the record.

Plate models have not been tested in trials showing that people who follow them get better health outcomes. They are communication tools drawn from expert consensus, not evaluated interventions. Their value is educational.

The nutrient requirements quoted come from the NHS because they are published in a numeric, accessible and verifiable format. Swiss and European reference values exist and sometimes differ slightly. These figures give an order of magnitude, not an individual prescription.

Finally, no plate model replaces professional advice in pregnancy, illness, a strict vegan diet in a child, or suspected deficiency. A blood test measures what a visual fraction never will.

Frequently asked questions

Which proportions should I use if I cook in Switzerland? Two fifths fruit and vegetables, two fifths starchy foods, one fifth protein-rich foods. That is the FSVO balanced plate, published in 2025 with the Swiss Society for Nutrition and Health Promotion Switzerland.

Do legumes count as protein or as starch? Both, and Swiss guidance says so explicitly. Legumes contain starch as well as a lot of protein, which makes them a valid alternative to other starchy foods. In practice they fill part of each section, which is why a lentil dish needs less rice.

How many legumes per person per meal? 60 g dry is one official portion, roughly 150 g cooked. To move towards the protein profile seen in the 2026 DIAAS study, aiming for about twice as much legume as grain by dry weight is a useful marker.

Do I really have to eat the vegetables first? The available evidence is favourable but limited: acute effects on blood glucose measured in samples of 18 to 107 young, healthy participants. The move costs nothing and requires no recipe change, which is enough to make it reasonable.

How do I apply the model to a soup or a curry? By looking at the recipe rather than the plate. Check the quantity of each group that went into the pot, then top up if needed with a salad, bread or a portion of cheese. That is the method the FSVO FAQ recommends.

Can a vegetarian skip supplements? An ovo-lacto vegetarian eating eggs and dairy regularly generally covers B12 and iodine through food. A vegan must supplement B12 without exception and keep an eye on iodine. Vitamin D concerns everyone in winter, whatever the diet.

Do the proportions change if I eat four times a day? The model describes a typical main meal. The NHS points out that balance plays out over a day or a week rather than at each eating occasion. A snack of fruit and seeds does not need to reproduce the full plate.

Sources