Substance + claimed effects

The substance is the preparation choice for vegetables — raw versus cooked, and within cooked, the method (steam, boil, microwave, sauté, roast). Heat changes vegetables along three independent axes: it inactivates plant enzymes (notably myrosinase in cruciferous), destroys heat-labile vitamins (vitamin C, folate, thiamine), and softens cell walls so that fat-soluble pigments (lycopene, β-carotene, lutein) and bound polyphenols become more extractable in the gut. Water-based methods also leach water-soluble micronutrients and antinutrients (oxalate, nitrate, some glucosinolates) into the cooking liquid. The article covers the bioavailability consequences (vitamins, minerals, carotenoids, glucosinolates/sulforaphane, polyphenols), the digestibility and satiety consequences, the meal-variety / dietary-adherence consequences, and the modest downstream meta scores (`health_short_term`, `beauty_cumulative`, `longevity`, with a minor `beauty_direct` lift via carotenoid-skin status). The article does not cover whether to eat vegetables at all — that is a separate entry; the marginal effect of preparation method is at least an order of magnitude smaller than the categorical effect of intake itself Aune et al. 2017.

Evidence by addressing question

Mechanism

Three mechanisms do almost all the work.

Heat destroys heat-labile vitamins. Vitamin C (ascorbic acid), folate, and thiamine degrade with time-at-temperature and oxidise on cut surfaces. In broccoli florets, conventional boiling cuts total glucosinolates by ~55% and vitamin C by a comparable amount, predominantly through leaching into the cooking water rather than thermal destruction; microwaving in water loses ~40% of vitamin C and ~74% of total glucosinolates; steaming preserves both with minimal loss Vallejo et al. 2002. The pattern repeats across vegetables: water-immersion methods (boil, pressure-cook) lose the most water-soluble nutrients; dry-heat methods (steam, microwave without water, griddle, roast) lose the least Jiménez-Monreal et al. 2009.

Heat denatures plant enzymes. The headline case is myrosinase in cruciferous vegetables (broccoli, cabbage, Brussels sprouts, kale). Myrosinase hydrolyses the glucosinolate glucoraphanin to sulforaphane, the isothiocyanate behind cruciferous vegetables' anti-cancer signal. Myrosinase denatures at ~60 °C; once it's gone, the gut microbiota can still produce some sulforaphane, but much less efficiently. In an 8-subject crossover trial, sulforaphane bioavailability from 200 g of crushed raw broccoli was 37% of the administered dose; from cooked broccoli, 3.4% — an order-of-magnitude difference — with peak plasma time of 1.6 h for raw versus 6 h for cooked Vermeulen et al. 2008. A 12-subject crossover by Conaway et al. 2000 found bioavailability of isothiocyanates from raw broccoli ~3× higher than from steamed broccoli where myrosinase was inactivated. The "chew raw broccoli or steam-only-briefly" advice is mechanistically well-grounded.

Heat softens cell walls and isomerises carotenoids. Carotenoids (lycopene, β-carotene, lutein, α-carotene) live inside chromoplasts surrounded by tough cellulose cell walls. Raw vegetables present these pigments in a matrix the small intestine has trouble extracting; cooking ruptures the cells and releases the pigments into the fatty phase of the meal, where mixed micelles can carry them to the enterocyte. Lycopene additionally isomerises from the trans- to the more bioavailable cis-form under heat. In humans, lycopene bioavailability from a 23 mg dose was 2.5–3.8× higher when delivered as tomato paste plus corn oil than as fresh sliced tomato plus corn oil Gärtner et al. 1997. In a 4-week crossover, plasma β-carotene response to 9.3 mg/d from processed (cooked, pureed) carrots and spinach averaged ~3× the response to the same dose from raw vegetables Rock et al. 1998. Cooking tomatoes at 88 °C for 30 min raised trans-lycopene content 2.6-fold and total antioxidant activity 1.6-fold even as vitamin C dropped ~30% Dewanto et al. 2002.

Evidence

The bioavailability literature is unusually clean — small mechanistic human trials with stable-isotope or plasma-response endpoints, replicated across labs, with consistent direction of effect by nutrient class. Beyond the trials already cited:

• Cooked tomatoes / carotenoid-rich vegetables ≫ raw. The Gärtner / Stahl / Sies line of work, replicated by Dewanto and others, is the strongest dataset on a single nutrient: 2–3× bioavailability advantage, robust to delivery vehicle, robust to dose Gärtner et al. 1997 Dewanto et al. 2002. The German raw-food-diet cohort makes the population-level case: long-term strict raw eaters had favourable plasma β-carotene but markedly low plasma lycopene compared with mixed-diet controls — exactly the gradient mechanism predicts Garcia et al. 2008.
• Raw or briefly-steamed cruciferous ≫ heavily cooked. Vermeulen 2008 (37% vs 3.4% sulforaphane bioavailability) and Conaway 2000 (~3× advantage for raw) are the human anchor trials; the in vitro and ex vivo myrosinase-denaturation curves match Vermeulen et al. 2008 Conaway et al. 2000. Vallejo 2002 fixes the optimal cooking method when raw isn't on the table: steam, not boil, not microwave-in-water Vallejo et al. 2002.
• Method matters more than fact-of-cooking. Miglio's three-vegetable, three-method head-to-head (carrots, courgettes, broccoli; steam, boil, fry) found steaming preserved carotenoids and ascorbic acid the best, boiling next, frying worst for antioxidant compounds though best for texture retention Miglio et al. 2008. Jiménez-Monreal's 20-vegetable, 6-method survey found that across most vegetables, dry-heat methods (griddle, microwave, bake) preserved antioxidant activity better than wet methods (boil, pressure-cook); fry was intermediate; total antioxidant capacity sometimes increased after cooking due to matrix softening and release of bound polyphenols Jiménez-Monreal et al. 2009.
• Downstream clinical-outcome data is thin and indirect. No RCT has randomised humans to "all-raw" vs "all-cooked" vegetable diets and tracked cancer, mortality, or cardiovascular endpoints. The categorical evidence — vegetable intake reduces all-cause mortality and CVD risk in dose-response fashion up to ~800 g/d Aune et al. 2017 — does not distinguish preparation. The raw-food cross-section in Koebnick 2005 (n=201) found favourable LDL and triglycerides but low HDL and B12 deficiency; that is partly a raw-vegan-bundle confound (no animal foods, no fortified grains), not pure preparation method Koebnick et al. 2005.

Protocol

The operational implication is variety of methods, matched to vegetable — not a single rule.

• Carotenoid-rich vegetables (tomatoes, carrots, sweet potato, pumpkin, red bell pepper, spinach, kale): cook with fat. Roast, sauté, or simmer in oil-containing sauces. The 2–3× bioavailability lift is robust. Adding any fat source (olive oil, avocado, eggs) at the meal further boosts absorption.
• Cruciferous (broccoli, cauliflower, Brussels sprouts, cabbage, kale, bok choy): eat raw chopped/grated, or steam 3–5 minutes (florets still bright green, slight crunch). Avoid boiling (leaches glucosinolates) and long roasting (denatures myrosinase). If you must use a high-heat method, add a pinch of mustard powder or mustard seeds to the cooked dish — mustard seed carries exogenous myrosinase that can partially rescue sulforaphane production from heat-killed broccoli.
• Vitamin-C-rich vegetables (bell pepper, broccoli, Brussels sprouts): include some raw portion. A raw bell pepper or a side of slaw covers the day's vitamin C with margin to spare; cooked bell pepper still has plenty.
• High-oxalate greens (spinach, Swiss chard, beet greens): boil and discard the water if kidney-stone-prone. Boiling leaches 30–87% of soluble oxalates into the cooking water; steaming is less effective (~30%). For most readers without stone risk, this isn't worth optimising around.
• Allium (garlic, onion): chop or crush, rest 10 minutes before heating. Allows alliinase to convert alliin to allicin before the enzyme denatures.
• Starchy roots (potato, beet, parsnip): raw isn't really an option; cooking method affects glycemic response (boiled cool potato < fried < baked hot mashed) more than nutrient retention.
• Dry-heat (steam, microwave without immersion, griddle, roast) outperforms water-immersion (boil, pressure-boil) for most nutrient classes Jiménez-Monreal et al. 2009.
• If you boil, use the water. Soup, stock, sauce base. The water-soluble vitamins and minerals are not destroyed by cooking; they're relocated. Recovering the cooking liquid recovers most of the loss.

Misconceptions

• "Raw is always healthier." The strongest single piece of evidence against this is the 2–3× bioavailability disadvantage of raw vs cooked for the carotenoid family and the order-of-magnitude difference between raw and processed tomato for lycopene Gärtner et al. 1997 Dewanto et al. 2002 Rock et al. 1998. The German raw-food cohort, who ate 1.5–1.8 kg of raw plant food per day, still ended up with low plasma lycopene Garcia et al. 2008. Raw is better for some nutrients (vitamin C, sulforaphane via myrosinase), worse for others (lycopene, β-carotene, lutein), and roughly equal for fibre, minerals (if water is consumed), and phenolic acids.
• "Cooking destroys the nutrients." Mostly false for fat-soluble pigments and bound polyphenols; partly true for vitamin C, folate, thiamine, and (for cruciferous) sulforaphane precursors. Total antioxidant capacity often increases after cooking because matrix softening releases bound compounds Dewanto et al. 2002 Miglio et al. 2008 Jiménez-Monreal et al. 2009.
• "Microwaving is the worst." Opposite, usually. Microwaving without water immersion preserves antioxidants and vitamins comparably to steaming, sometimes better than boiling, because exposure time is short and there's no leaching medium Jiménez-Monreal et al. 2009. Microwaving cruciferous in standing water is the bad version Vallejo et al. 2002.
• "Cooked vegetables = lost minerals." Minerals don't degrade at cooking temperatures; they leach into water. Steam, dry-roast, or use the cooking liquid and the mineral budget is intact.
• "Fresh produce beats canned/frozen." Canned and frozen vegetables are typically blanched at peak ripeness; nutrient retention is comparable to or better than fresh produce that has spent a week in transit and a few days on the counter. Cooked canned tomatoes deliver more bioavailable lycopene than raw fresh tomatoes.

Practicalities

This dimension is what makes preparation method a real lever: it changes how many vegetables a person actually eats. Cooked vegetables are denser, smaller per serving, hot — a person can put down 400 g of stir-fry comfortably and would struggle to chew through the equivalent raw salad. Cooking unlocks meal templates (curries, stews, stir-fries, roasted pans, soups) that don't exist for raw. In practice, a strict-raw approach concentrates eaters around salads, smoothies, and crudités, which most people tire of within months; a flexible-method approach lets vegetables ride along with whatever the rest of the meal already is. Survey data on dietary adherence is not specific to raw-vs-cooked, but the raw-food literature consistently finds adherent populations are highly self-selected and dietary-motivation–dominant Koebnick et al. 2005 Garcia et al. 2008.

Digestibility is the second practical lever. Cell-wall rupture by cooking reduces the chewing and gastric work required, lowers the fermentable substrate that reaches the colon as bulk (some shifts from insoluble fibre to soluble/short-chain), and reduces gas/bloating for people with IBS, SIBO, or post-surgical guts. The pattern: people whose guts tolerate raw cruciferous fine are a minority report; most adults find lightly-cooked versions sit better. There is no clean RCT here; the consensus is clinical-pattern level.

Failure modes

• The "I only eat salad" trap. Vegetable intake plateaus at whatever a person can chew through cold; cooked-vegetable channels (soup, sauté, roast tray) are missing. Total vegetable intake ends up lower than a flexible-method eater achieves.
• The "boil everything" default. The single highest-loss method becomes the only method, and the cooking water gets poured down the drain. Half the water-soluble vitamins and minerals leave with it Vallejo et al. 2002 Miglio et al. 2008.
• The "no fat with vegetables" mistake. Carotenoid absorption is fat-dependent. Dressing a salad with vinegar alone, or eating roast vegetables without any oil, leaves most of the carotenoids in the gut lumen. Adding ~5–10 g of fat per carotenoid-containing meal is the cheap fix.
• Overcooking cruciferous. Boiling broccoli until grey is the canonical example: most glucosinolates leached, myrosinase denatured, sulforaphane production crippled Vallejo et al. 2002 Vermeulen et al. 2008.
• The "raw kale smoothie" superfood inflation. Raw cruciferous gives sulforaphane only with thorough chewing or mechanical disruption; throwing whole leaves into a blender without much shear delivers less than expected. Sprouts, mustard-seed addition, and brief steam-then-blend recover most of the gap.

Contraindications

Two narrow, real ones.

• Calcium-oxalate kidney-stone formers should boil-and-discard high-oxalate greens (spinach, Swiss chard, beet greens). Boiling cuts soluble oxalate 30–87% depending on time and water volume; steaming and microwaving leave most of it in the vegetable. This subgroup also benefits from pairing oxalate-rich vegetables with calcium sources, which precipitates oxalate in the gut.
• Iodine-deficient individuals with thyroid disease should be modest with large daily doses of raw cruciferous; glucosinolate-derived goitrins can modestly suppress iodine uptake. Population effect is small at typical intakes; clinically relevant in subsistence diets where cassava or raw brassicas dominate.

Food safety, gut-friendly preparation (cooking softens for IBS / post-surgical / dental issues), and immunocompromised raw-produce risks (listeria, E. coli on raw sprouts and leafy greens) are background concerns rather than preparation-specific contraindications.

Out of scope

The article points to: total vegetable intake (the dominant lever, separate entry); fibre and the gut microbiota; oxalates and kidney stones; fermented vegetables (a third pathway alongside raw/cooked); legumes and lectins/phytates (where cooking is non-optional and the safety case is decisive); cooking-oil selection; acrylamide in high-temperature starchy cooking.

Credibility range

Optimist case (raw is meaningfully better)

Heat-labile vitamins (C, folate, thiamine) are real and meaningful; cooking can cut vitamin C by 30–60% in water-based methods Vallejo et al. 2002. Myrosinase activity, the single most important enzyme system in cruciferous, is destroyed by cooking — and the cruciferous–cancer story rides on sulforaphane Vermeulen et al. 2008 Conaway et al. 2000. Long-term raw food cohorts show favourable lipid profiles (low LDL, low triglycerides), favourable plasma β-carotene, and protective changes in inflammatory markers Koebnick et al. 2005 Garcia et al. 2008. Raw food is more filling per calorie, supports higher-volume eating patterns associated with lower body weight, and avoids the formation of cooking-related carcinogens (acrylamide, AGEs, heterocyclic amines on starches and proteins respectively).

Skeptic case (the dichotomy is overstated)

The bioavailability advantage for cooked carotenoids is 2–3× and replicated across labs and across pigments; this is the single largest preparation-method effect in the human-trial literature Gärtner et al. 1997 Rock et al. 1998 Dewanto et al. 2002. The German raw-food cohort, eating 1.5–1.8 kg of raw plant food per day, ended up low in plasma lycopene — a real population-level signal that preparation matters at intakes where intake itself isn't limiting Garcia et al. 2008. Cooked vegetables are more digestible, kinder to a typical adult gut, and support higher total vegetable intake — which is by far the biggest lever Aune et al. 2017. No human RCT randomises raw vs cooked vegetable diets and measures hard outcomes; both sides operate on bioavailability proxies plus mechanism. The raw-only failure modes (B12 deficiency, low HDL, low lycopene) reflect dietary-bundle bias more than a categorical raw advantage Koebnick et al. 2005. The cooking-carcinogen concern (acrylamide, AGEs) lives almost entirely in high-temperature starchy and meat cooking, not in steamed broccoli or roasted carrots.

Author's call

The empirical answer is mixed-method, vegetable-by-vegetable. The bioavailability literature is asymmetric in a way that makes "cook most of them, eat some raw" the clearly correct projection: the carotenoid family wins big from cooking with fat (2–3×), water-soluble vitamins lose modestly from boiling (recoverable by steaming or using the cooking liquid), cruciferous sulforaphane wins big from raw or brief steam (3–10×). The right action is variety of method, matched to vegetable, with a default toward steam/sauté/roast and a deliberate raw component for vitamin C and cruciferous. The bigger lever, by an order of magnitude, is whether the person eats vegetables at all Aune et al. 2017; preparation method is the optimisation that sits on top. Meta scores reflect that secondary status — real but bounded.

Stakeholder + incentive map

• Raw-food / living-foods movement — long-standing health subculture, commercial ecosystem of dehydrators, raw cookbooks, retreat centres. Frames cooking as nutrient destruction, often conflates raw-veganism with raw-vegetable preparation. Genuine evidence on raw cruciferous and vitamin C, overreach on carotenoids and minerals.
• Mediterranean / traditional cuisine advocacy — restaurant industry, olive-oil producers, popular nutrition writers (Pollan, Bittman). Frames cooking-with-fat as the natural preparation. Aligned with the carotenoid-bioavailability literature.
• Processed food industry (canned/frozen tomato, vegetable oil) — has a real commercial interest in the "cooked is better" framing, which happens to be correct for tomatoes and most carotenoid vegetables. Funded a good chunk of the lycopene bioavailability work; findings replicate in independent labs.
• Cancer-prevention research community — Johns Hopkins (Talalay) and the cruciferous-vegetable line pushes raw-or-briefly-steamed for sulforaphane. Aligned with mechanism, somewhat at odds with the cooked-tomato camp on framing but not on substance.
• Wellness / juicing influencer ecosystem — commercial alignment with raw smoothies, juicers, cold-pressed juices. The economics push raw; the science is mixed and overstated.
• Public-health / dietary-guideline bodies (USDA, EU EFSA, WHO) — agnostic on preparation, push total intake. Aligned with the Aune meta-analysis: dose matters far more than method.

Population variability

• Baseline carotenoid status / fat intake — readers who eat very low-fat (raw-vegan, ultra-low-fat) gain the most from adding fat to carotenoid-rich vegetables; readers already eating Mediterranean-style with olive oil at most meals have the bioavailability bonus baked in.
• Gut sensitivity — IBS, SIBO, post-surgical, elderly, and dental-issue populations gain markedly more from cooked over raw; raw cruciferous and onion-family in particular drive bloating in these subgroups. Healthy young guts handle raw fine.
• Vitamin A / pro-vitamin A status — populations relying on plant-based vitamin A (vegetarian / vegan, low-income, developing-world) gain disproportionately from cooked carotenoid-rich vegetables; populations with regular retinol intake from animal foods are insulated.
• Kidney-stone history — narrow subgroup where boil-and-discard high-oxalate greens is meaningful.
• Thyroid / iodine status — iodine-deficient individuals modestly affected by raw cruciferous goitrins; iodine-sufficient populations effectively unaffected at culinary intakes.
• Cultural baseline — South Asian, Mediterranean, East Asian cuisines are pre-optimised for cooking with fat / brief stir-fry; Northern European and Anglo-American defaults to overcooked vegetables underperform. Salad-heavy Levantine / Tex-Mex patterns get vitamin C and chewing volume but under-deliver carotenoids without dressing fat.

Knowledge gaps

• No RCT randomises humans to all-raw vs mixed-method vegetable diets with hard outcomes. Everything sits on bioavailability proxies and observational cohorts of self-selected raw eaters.
• Microbiome conversion of glucosinolates to sulforaphane in cooked-broccoli eaters varies dramatically by individual (likely by gut microbiota composition), and is not well-mapped — the average is 3.4% bioavailability but the spread is wide.
• The dose-response curve for "how much fat is enough" with carotenoid meals — most evidence points to ~3–5 g of fat being sufficient, but the curve is poorly mapped at the low end.
• Long-term outcomes of method-mixing versus single-method patterns are not separable from cuisine and dietary-pattern confounds.
• The interaction between preparation method and ultra-processed food displacement — most preparation-method studies are done on the vegetable in isolation, not on the meal-pattern shift that adopting them creates.

Scope held narrow on purpose. The brief named "bioavailability of vitamins, minerals, and phytochemicals, digestibility, satiety, and meal variety" — all covered, with bioavailability getting the most weight because it's where the human-trial evidence is cleanest and where the magnitude is largest. Satiety / volume-per-calorie got a short treatment under practicalities and failure-modes rather than its own section; the satiety effect is real (raw is more filling per calorie) but is a second-order consequence of the preparation choice, and devoting a section to it would have crowded out the bioavailability material that's the article's actual punch.

Excluded by design:

• Whether to eat vegetables at all — different substance, dominant effect, belongs in its own entry. Article explicitly defers to that.
• Fermented vegetables — flagged as a separate adjacent pathway in out-of-scope; not a raw/cooked subcase.
• Legume preparation (cooking, soaking, lectins) — different safety case (cooking non-optional), warrants own entry.
• Acrylamide and cooking carcinogens — lives almost entirely in high-temperature starchy/protein cooking, not in vegetable preparation as scoped. Mentioned in research §3c (skeptic case) but kept out of the article body to avoid confusing the carotenoid-cooking message.
• Specific micronutrient retention tables per vegetable per method — would have turned the article into a reference manual; the protocol section's by-vegetable rules cover the action-relevant 80%.

Rating calls worth flagging:

• longevity = 1 not 0 — borderline. The categorical vegetable-intake effect on mortality is huge Aune 2017; the marginal contribution of method is unmeasured but mechanistically plausible (bioavailable carotenoids, preserved sulforaphane). 1 reflects "marginal contribution" honestly without inflating.
• health_short_term = 2 not 1 — digestibility and gut-tolerance benefits are clinical-pattern-level not RCT-level, but the felt-day-to-day improvement when a sensitive eater switches to mostly-cooked is real. 2 fits "small but real daily-feel improvement."
• beauty_direct = 1 not 2 — kept low because the carotenodermia signal is subtle and needs both regular intake and the bioavailability lift to land visibly. Could go to 2 if reviewers feel the carotenoid-skin literature is stronger.
• action = do not know — borderline. This is more "decision literacy applied across many small choices" than a single habit, which fits know; but the protocol section has concrete by-vegetable rules a reader can actually adopt, which is what do requires. Settled on do.

Future-link candidates: fiber-intake, fermented-foods, olive-oil, fruit-vegetable-intake, legumes-and-lectins, acrylamide-cooking-temperatures, kidney-stones-oxalate, cruciferous-vegetables-cancer, broccoli-sprouts-sulforaphane. The last three are strong separate-entry candidates whose detail would otherwise creep into this one.

Contraindications field deliberately empty in meta. The closed vocabulary doesn't include kidney-stone history as a token, only the more general kidney-disease — which is broader than the specific oxalate-stone subgroup the contraindications section addresses. Reflected that subgroup-narrowness in prose instead.