Which Microgreen Is Most Nutritious? Nobody Has Established One, and the Beta-Carotene Podium Is 15 Micrograms Wide

There is no published, matched-conditions, per-serving comparison that supports naming one microgreen the most nutritious. A crop can lead a specified compound in a specified experiment, and the leader changes with the compound, the cultivar, the light, the substrate, the harvest stage, the water basis and the amount you actually eat.

That is not a dodge, and this page is not going to stop there. The interesting part is what happens when you take the best available dataset and convert it into the serving a person actually eats. The ranking does not survive the conversion.

In one study, four different crops led four different compounds

The broadest dataset available measured 25 commercially grown named varieties for total, free and dehydro ascorbic acid, beta-carotene, lutein and zeaxanthin, violaxanthin, phylloquinone and alpha- and gamma-tocopherol, all reported per 100 g fresh weight.

No crop led vitamin C, K1, beta-carotene and the tocopherols together. Cilantro led two of the three named carotenoid measures and lost the third.

One correction worth making explicitly. Summaries of this study often name cilantro the carotenoid leader. Red sorrel led beta-carotene, at 12.1 against cilantro's 11.7 and red cabbage's 11.5. Cilantro's wins were lutein and zeaxanthin, and violaxanthin.

And a second study finds different leaders again. Among four hydroponic brassicas measured under one protocol, kale was highest in total polyphenols, mustard and radish highest in total isothiocyanates, and mustard highest in total carotenoids on that study's dry-weight basis. Different panel, different leaders.

Convert it to a real serving and the ranking collapses

The whole table above is written against 100 grams of fresh microgreens. Nobody eats that.

The only weighed serving anybody publishes is 2.5 grams a person, which is one fortieth of the reference quantity the entire literature is denominated in. Convert the leaders and the picture changes completely.

The entire distance between the best and worst crop in that study for vitamin C, at a real serving, is 3.17 mg.

And the beta-carotene podium is the number that ends the argument. First place was 12.1 mg per 100 g and third place was 11.5. At a 2.5 g serving that is 0.3025 mg against 0.2875 mg.

Fifteen micrograms separates first from third.

Those three crops are 5.2% apart, which is a gap you could produce by harvesting a day later, using a different lamp, or picking a different cultivar of the same crop. It is not a ranking. It is noise with an order imposed on it.

The fresh-weight basis is partly a ranking of which crop holds less water

This is the confound sitting inside that same table, and it is much larger than the gaps it is being used to decide.

The 25 crops ranged from 4.6% to 10.2% dry matter, mean 6.9%. So 100 g of the driest crop contains 10.2 g of actual plant and 100 g of the wettest contains 4.6 g.

A per-fresh-weight ranking rewards the drier crop, because more of its 100 grams is plant. The moisture spread in this dataset is roughly 23 times the gap it is being used to settle. Any ordering where crops finish within a few percent of each other is being decided by water content before it is decided by nutrition.

You cannot fix this by converting with a generic moisture figure. Conversion needs each sample's own dry-matter fraction: concentration per gram dry weight equals concentration per gram fresh weight divided by the dry-matter fraction. The study publishes enough dry-matter data to do that for itself, and it does not publish a recalculated ranking on a dry basis.

Mixing bases across studies is worse. The four-brassica study reports vitamin C on fresh weight and total carotenoids in beta-carotene equivalents per 100 g dry weight, in the same paper. Setting its carotenoid figures beside the 25-crop study's carotenoid figures, as though they shared a basis, is not a comparison.

The conditions were not matched, so the crop is not the only variable

Calling this a crop ranking requires the crops to be the thing that differed. In the 25-crop study they were not.

• 23 crops were soil-grown in an unheated, ambient-light greenhouse.
• China rose and green daikon radish were hydroponic.
• Golden pea tendrils and popcorn were dark-grown.
• Harvest age varied by species, and all of it came from one grower and one analytical lab.

The authors point at the cleanest illustration themselves. Golden pea tendrils and green pea tendrils were the same seed source, one grown in light and one in darkness, and the dark-grown sample was among the low carotenoid and K1 results. That is a lighting result presented in a crop table, and it is a direct warning against reading any row as a property of the plant.

So the honest wording is narrow: under these stated commercial conditions, this sample was higher for this analyte. Not: this microgreen is healthier.

Cultivar can reverse the answer inside a single crop

"Radish" is not a specification, and a study that held everything else constant proves it.

Five radish cultivars, tray-grown without substrate under artificial light and harvested at day 10:

• 'Asia red' had the highest vitamin C, at 236.27 mg per 100 g dry weight.
• 'Koregon red' had the highest total phenolics, at 280.53 mg per 100 g as that study reported it, plus high flavonoids.
• Glucoraphasatin and glucoraphanin proportions differed sharply among the five.
• Dry matter ran 4.75% to 7.65% across cultivars of one crop, which is most of the spread the 25-crop study found across 25 different crops.

A separate three-cultivar experiment reached the same place from the light side, finding that cultivar and LED type both moved phenolics, ascorbic acid, pigments and antioxidant capacity.

A page that tells you to buy "radish" has not told you which of these you are getting. Meanwhile a controlled four-brassica study found genotype was the principal source of variation, with harvest-stage effects limited and themselves genotype-dependent: beta-carotene showed no stage response while lutein and ascorbate responses varied by genotype.

Bioaccessibility is not absorption, and it varies more than the concentrations do

Concentration in the leaf is not the amount that reaches you.

The four-brassica study ran simulated gastrointestinal digestion and found sharply different accessible fractions:

Mustard's accessible vitamin C fraction is 3.4 times broccoli's, and radish's accessible isothiocyanate fraction is roughly double the other three. Those spreads are far larger than the concentration gaps the ranking pages argue over.

But this is a digestion model, not people. It measures bioaccessibility, what is released and potentially available, not bioavailability, what a human actually absorbs and uses. No study located measures human absorption by microgreen crop. Settling that needs randomized feeding studies with characterized batches and blood or urine biomarkers.

What the ranking pages actually supply

Four "healthiest microgreens" pages were opened and audited on 2026-08-11.

Zero of four supplied an auditable matched study supporting the ordering itself, at the point where the ordering was claimed. One opened with arugula, one gave a top-ten-style guide and the "up to 40 times" wording, one listed seven "healthiest" crops in a home-growing guide, and one named broccoli "king of nutrients" with a comparison table lacking the conditions needed to compare its own inputs.

None of them published, next to its ranking, the cultivar, harvest stage, growth conditions, fresh or dry basis, a common serving, and the calculation converting measurements into that order. That is four pages, chosen and dated, not a census of the web.

Broccoli is the most common explicit number one, usually justified with general brassica reputation and glucosinolate language rather than a matched cross-crop per-serving score. That describes what gets published. It is not evidence that broccoli wins.

The "up to 40 times" line does its usual work here too. It is a maximum, analyte-specific and comparator-specific statement about microgreens against mature counterparts, and it is used to support a healthiest-microgreen message it cannot address at all, because it says nothing about how microgreens rank against each other. The dedicated page takes that number apart.

A second recurring error is treating microgreens and sprouts as interchangeable, which FDA's rule explicitly does not, and which part one of this series covers in full.

What to actually do

What nobody has measured

• A matched multi-crop panel: one named cultivar set, grown and harvested together, comprehensive validated assays reported on both fresh and dry weight, with a pre-specified scoring rule declared before the data is seen.
• A comprehensive compound panel. The available datasets do not jointly cover B vitamins, protein and amino acid quality, fatty acids, fiber, sugars, sodium, oxalate, phytate, nitrate and contaminants across one representative crop set. Absence from a panel is not absence from the plant.
• A per-serving cross-crop comparison using a consumer-tested serving mass and equal fresh-weight servings, including handling losses.
• Human bioavailability by crop. Everything published is in-vitro digestion.
• Whether a mix beats a single crop at equal total serving mass, for delivery, acceptability and absorption.
• A census of the unmeasured crops. Sunflower, wheatgrass, chive and onion, fennel and most branded mixes have no crop-specific matched multi-nutrient profile. That is a gap in the literature, and it is not a finding about those crops.

Sources

Opened 2026-08-11. The serving conversions are this site's own arithmetic: each published per-100-g fresh-weight concentration multiplied by 2.5 g and divided by 100, using the 2.5 g serving this site already publishes as the only weighed serving in circulation. The vitamin K1 row converts 4.1 µg per g to 410 µg per 100 g first. The 23-fold confound figure divides the dataset's own dry-matter spread by its own beta-carotene podium gap. None of these conversions introduces an outside number, and none of them is a recommended intake.

• Xiao, Lester, Luo and Wang, 2012, vitamin and carotenoid concentrations of edible microgreens (PDF) - 25 commercially grown named varieties at cotyledon or very early true-leaf stage, reported per 100 g fresh weight: red cabbage leading total ascorbic acid at 147.0 mg, red sorrel leading beta-carotene at 12.1 with cilantro at 11.7 and red cabbage at 11.5, cilantro leading lutein and zeaxanthin at 10.1 and violaxanthin at 7.7, garnet amaranth leading phylloquinone at 4.1 µg per g, and green daikon radish leading alpha- and gamma-tocopherol at 87.4 and 39.4 mg; the lowest reported total ascorbic acid of 20.4 mg per 100 g; dry matter of 4.6 to 10.2% with a mean of 6.9%; and the methods showing 23 crops soil-grown in an unheated ambient-light greenhouse, China rose and green daikon radish hydroponic, golden pea tendrils and popcorn dark-grown, and the golden and green pea tendrils sharing a seed source under dark and light. This is the same 2012 study cited on the baby greens comparison and on the 40 times page, reached there through a different hosted copy.
• de la Fuente et al., 2019, four hydroponic Brassicaceae microgreens - broccoli, curly kale, red mustard and radish under one protocol, measuring minerals, vitamin C, total carotenoids, polyphenols, anthocyanins, isothiocyanates and simulated-digestion bioaccessibility; kale highest in total polyphenols, mustard and radish highest in total isothiocyanates, mustard highest in total carotenoids on that study's dry-weight basis; vitamin C bioaccessibility of 1.10% broccoli, 1.87% kale, 3.73% mustard and 2.61% radish, and total isothiocyanate bioaccessibility of 32.30%, 34.06%, 31.07% and 63.36% respectively; and the mixed reporting, with vitamin C on fresh weight and total carotenoids per 100 g dry weight.
• Kim et al., 2023, five radish cultivars - 'Asia green 1', 'Asia green 2', 'Asia red', 'Koregon red' and 'Asia purple' tray-grown without substrate under artificial light and harvested at 10 days: 'Asia red' highest in vitamin C at 236.27 mg per 100 g dry weight, 'Koregon red' highest in total phenolics at 280.53 mg per 100 g with high flavonoids, sharply differing glucoraphasatin and glucoraphanin proportions, and dry matter ranging 4.75 to 7.65%.
• Kyriacou et al., 2021, four Brassicaceae types at two stages - komatsuna, mibuna, mizuna and pak choi under controlled conditions, measuring yield, minerals, carotenoids, phenolics and anthocyanins, ascorbate and antioxidant capacity; genotype as the principal source of variation, with first against second true-leaf effects limited and genotype-dependent, beta-carotene showing no stage response, and lutein and ascorbate responses varying by genotype.
• Johnson et al., 2020 and 2021, six species against mature counterparts - arugula, broccoli, red cabbage, red beet, red amaranth and pea compared by non-targeted metabolomics and ionomics across 26 minerals, demonstrating different profiles rather than an overall winner.
• Three radish cultivars under two LED spectra - grown at 24 °C on a 16 hour and 8 hour photoperiod, concluding that cultivar and LED type both affect phenolics, ascorbic acid, pigments and antioxidant capacity.
• Six-species comparison, mungbean, lentil, red radish, pearl millet, mustard and red cabbage - phenolics, anthocyanins, vitamin C, fiber, phytic acid and DPPH activity, useful for a narrow assay comparison rather than an overall nutrient score.
• FDA, guidance for industry on sprouts (PDF) - the distinction between substrate-grown microgreens cut above the substrate and sprouts, for sprout-specific Produce Safety Rule coverage.
• Practice sample, four ranking pages opened and audited 2026-08-11 as a record of what is published rather than as evidence, and the source of the zero-of-four count: The Wonderful World of Sprouts, which opens its order with arugula; Microgreens World, which repeats "up to 40 times" alongside a generic rank; Clean Eating Kitchen, a seven-crop home-growing list labeled "best"; and Total Gardener, which names broccoli "king of nutrients" and presents a comparison table without the conditions needed to compare its inputs.