How many vegetables do you really need per day?
Five servings of vegetables and fruit per day offers the best protection, with non-starchy vegetables counting and juice and potatoes not. More than five servings adds nothing.
Read the full answer
Five servings of vegetables and fruit per day is the number with the strongest evidence base. This comes from a meta-analysis of 26 cohort studies involving nearly 1.9 million participants, supplemented by two large American cohorts of more than 108,000 people followed over thirty years. People who reach five servings have a 13% lower chance of dying than those who eat only two. More than five servings per day adds nothing extra.
The ratio that performs best is roughly two servings of fruit and three servings of vegetables. For your heart, that translates to a 12% lower risk of dying from cardiovascular disease. The most striking effect is for lung disease: the risk of dying from it was no less than 35% lower at five servings compared with two.
Not all vegetables carry equal weight. Fruit juice and potatoes showed no protective effect in that same analysis. Starchy vegetables such as peas and corn did not either. Leafy greens and non-starchy vegetables are the best choice. That immediately gives you a practical guideline: focus on varied, non-starchy vegetables, and skip the juice.
A major stumbling block is that most people think they are already eating enough vegetables, when in fact they are not. In the United States in 2019, only 1 in 10 adults reached the recommended amount. In Australian research, more than half of the people who fell short thought they were doing fine. This self-overestimation is the main reason why many people never adjust their habits.
There are also indications that eating a lot of vegetables is associated with less cognitive decline in later life, but that evidence is preliminary. It comes from a single cross-sectional study in a specific Brazilian population, and demonstrating an association is different from proving a cause. That finding is therefore interesting but not yet sufficient to make firm statements about it.
All mortality data come from PMID 33641343 (large meta-analysis + two cohorts). US consumption data: PMID 34990439, Australia: PMID 35491549. Nutrient balance: PMID 38572364. Cognition: PMID 24736378 (cross-sectional, limited). Small RCT gestational diabetes: PMID 37630814 (too small for firm conclusions, combination intervention). All associations are observational; no randomised experiments exist for the mortality figures.