Croissants: the ones in the studies were not the ones in the bakery

Nearly everything research knows about the croissant was learned from croissants built for the experiment. A sourdough croissant reformulated for that trial to carry 4.8 g of fibre per 100 g lowered fasting blood glucose by 2.0 mg/dL over two weeks, while the plain croissant in the same trial raised it by 3.1 mg/dL. In those same 32 people nothing else moved: appetite, energy intake, blood lipids, insulin, C-reactive protein and the gut microbiome all stayed where they were. Croissants and muffins carrying 3.2 g of plant sterol esters a day took LDL cholesterol down 14.7 percent against a plain bakery control over eight weeks, and the sterols were the thing under test. The pastry you buy is the control arm in this research, and the control arm is the one that moved nothing. One more thing belongs up here rather than in small print at the end. The only safety evidence we hold about croissants concerns two HIV medicines: whether you eat a croissant with them decides how much of the dose reaches you. Past that there is nothing. No upper amount, nothing on pregnancy, children or kidney disease, no other drug interaction. Read that silence as silence, not as a clean bill of health.

The numbers, per 100 g

What cooking and storage change

• A sourdough croissant reformulated to 4.8 g fibre per 100 g lowered fasting blood glucose by 2.0 mg/dL over 2 weeks, while the control croissant raised it by 3.1 mg/dL. — 32 healthy adults, 2-week breakfast intervention, parallel groups (n = 32) Evidence B source
  Fasting glucose change from baseline -2.0 mg/dL vs +3.1 mg/dL, P = 0.022
• Two sourdough croissants emptied from the stomach faster and produced 30 percent less breath hydrogen than two brewer's yeast croissants, and postprandial symptoms were 36 percent milder over 4 hours. — 17 healthy adults aged 18 to 40, double-blind randomised crossover, single meal of 2 croissants (n = 17) Evidence B source
  Gastric volume AUC 11 percent lower over 3 h (P = 0.02), hydrogen production 30 percent lower over 4 h (P = 0.03), symptom severity 36 percent lower (P = 0.05)

What it does to your health

• Croissants of 1.80 MJ made with a fat substitute and ordinary croissants of 3.45 MJ left the same energy intake afterwards in 34 normal-weight adults, so the lighter croissant bought no saving across the day. — 16 men and 18 women of normal weight, six experimental conditions, ad libitum intake after 0.25, 2.25 and 4.75 hours (n = 34) Evidence B source
  No difference in subsequent energy intake between the fat-substitute and the high-fat croissant
• Croissants and muffins carrying 3.2 g of plant sterol esters a day for 8 weeks lowered LDL cholesterol by 0.26 mmol/L in people with normal cholesterol, a 14.7 percent difference against the plain bakery control. — 57 normocholesterolemic men and women, 29 control and 28 sterol ester group, two bakery pieces daily for 8 weeks (n = 57) Evidence B source
  LDL cholesterol down 0.26 mmol/L within the sterol ester group (P < 0.005); between-group difference 0.36 mmol/L or 14.7 percent (P < 0.001)

Safety, limits and interactions

• Taking etravirine 100 mg with a croissant breakfast gave 80 percent of the exposure seen after a standard breakfast, while taking it fasted gave only 49 percent. — 24 HIV-negative healthy male volunteers in two panels, phase I randomised crossover, single 100 mg dose (n = 24) Evidence B source
  AUClast ratio 0.80 (90% CI 0.69 to 0.94) for the croissant breakfast and 0.49 (90% CI 0.39 to 0.61) fasted, both vs standard breakfast
• A croissant with coffee worked as well as a standard or high-fat breakfast for darunavir absorption, while fasting cut exposure by about 30 percent. — HIV-negative healthy volunteers, open-label randomised crossover, single 400 mg darunavir dose with ritonavir 100 mg twice daily Evidence B source
  Cmax and AUClast about 30 percent lower fasted than after a standard meal; no significant difference between fed states

What people believe that the data does not support

• In a Spanish cohort of 8964 adults followed for a median 6.2 years, commercial baked goods such as croissants showed no dose-response link with depression, although any consumption above the lowest fifth carried a hazard ratio of 1.38. — 8964 adults in the SUN cohort, Spain, median follow-up 6.2 years, 493 incident cases of depression (n = 8964) Evidence C source
  No linear relationship for commercial baked goods; quintiles 2 to 5 vs quintile 1 hazard ratio 1.38 (95% CI 1.06 to 1.80)

Who this works differently for

• In 26 people with non-celiac gluten sensitivity, an oral provocation test with gluten-containing croissants produced symptoms no more severe than the same test with gluten-free croissants. — 26 of 60 patients with non-celiac gluten sensitivity undergoing an oral provocation test, against 40 celiac patients and 20 healthy volunteers (n = 26) Evidence C source
  No significant difference in symptom severity between gluten-containing and gluten-free croissants

What is still unknown

• In the same 2-week trial nothing else moved: appetite, energy intake, blood lipids, insulin, C-reactive protein and the gut microbiome were all unchanged by the fibre-enriched croissant. — 32 healthy adults, 7-day weighed food diaries, blood, urine and faecal sampling (n = 32) Evidence B source
  No change detected in any other monitored variable or in the gut microbiome

The bottom line

Two findings here you can act on. If croissants sit heavily on you, the leavening may be why. Seventeen healthy adults who ate two sourdough croissants emptied their stomachs faster, produced 30 percent less breath hydrogen and rated their symptoms 36 percent milder over four hours than after two croissants made with brewer's yeast. That was one meal in healthy people under 40, so treat it as something to try rather than a rule. The second is that the lighter croissant buys you nothing. Croissants made with a fat substitute at 1.80 MJ and ordinary ones at 3.45 MJ left 34 adults of normal weight eating the same amount later in the day, so what was saved at breakfast came back before evening. If you take etravirine or darunavir for HIV, food decides how much of the dose reaches you, and a croissant counts as food. A croissant breakfast gave 80 percent of the exposure a standard breakfast gave for etravirine, and taking it fasted gave 49 percent. With darunavir, a croissant with coffee matched a standard breakfast, and fasting cost about 30 percent. Which breakfast fits your regimen is a question for whoever prescribed it, and this page cannot answer it for you. Two common beliefs came out weaker than they sound. In 26 people with non-celiac gluten sensitivity, an oral provocation test with gluten-containing croissants produced symptoms no worse than the same test with gluten-free ones, which is 26 people in one test and no more than that. The people with celiac disease in that study were a comparison group and never took that test, so nothing here speaks to celiac disease. And in 8964 Spanish adults followed for a median of 6.2 years, eating any commercial baked goods at all carried a 38 percent higher rate of depression than eating almost none, while eating more of them did not raise the rate any further. That missing gradient is the reason nobody can call the pastry a cause here, and an observational study cannot say what else in those lives is doing the work. The biggest hole on this page has no card in it. Beyond the two HIV medicines above we found nothing on pregnancy, children, kidney disease or any other drug interaction, and no upper amount at all. Silence in our sources is not a clean bill of health.

Sources

1. crs-r3-01 · randomised controlled trial, 32 healthy participants
   FIBCRO consumption improved fasting blood glucose compared with CONCRO (mean changes from baseline -2.0 mg/dL in FIBCRO compared with +3.1 mg/dL in CONCRO, P = 0.022), also reducing serum DPPIV activity by 1.7 IU/L (P = 0.01) and increasing urinary excretion of urolithin A-sulfate by 6.9 ng/mg creatinine (P = 0.04) compared with baseline.
   J Nutr, 2024 · checked 2026-09-23
2. crs-r3-03 · double-blind randomised crossover trial, 17 healthy adults
   The total GV AUC was reduced by 11% during the 3 h after the consumption of SCs compared with BCs (P = 0.02). Hydrogen production during the 4-h interval after ingestion of SCs was 30% lower than after BCs (P = 0.03).
   J Nutr, 2018 · checked 2026-09-23
3. crs-r3-04 · controlled clinical trial, 34 adults
   The SPE croissants (1.80 MJ) and the high-fat croissants (3.45 MJ) did not result in different subsequent energy intakes.
   Physiol Behav, 1995 · checked 2026-09-23
4. crs-r3-07 · randomised double-blind placebo-controlled trial, 57 adults
   Total and LDL cholesterol (LDL-C) decreased in the SE group by 0.24 mmol/L (P < 0.01) and 0.26 mmol/L (P < 0.005), respectively, whereas these variables did not change in the control group.
   J Nutr, 2003 · checked 2026-09-23
5. crs-r3-05 · phase I randomised crossover trial, 24 healthy men
   When dosing occurred after a light or enhanced-fiber breakfast, the corresponding values were 0.80 (90% CI 0.69-0.94) and 0.75 (90% CI 0.63-0.90), respectively.
   Pharmacotherapy, 2008 · checked 2026-09-23
6. crs-r3-06 · open-label randomised crossover trial in healthy volunteers
   Administration of darunavir/rtv in a fasting state resulted in a decrease in darunavir C(max) and AUC(last) of approximately 30% compared with administration after a standard meal. No significant differences in darunavir plasma concentrations were observed between different fed states.
   J Clin Pharmacol, 2007 · checked 2026-09-23
7. crs-r3-08 · prospective cohort, 8964 adults
   No linear relationship was found between the consumption of commercial baked goods and depression. Participants belonging to consumption quintiles Q2-Q5 showed an increased risk of depression compared with those belonging to the lowest level of consumption (Q1; HR = 1·38; 95 % CI 1·06, 1·80).
   Public Health Nutr, 2012 · checked 2026-09-23
8. crs-r3-09 · case-control study, 26 provocation tests
   No significant difference was found between the severity of symptoms reported by NCGS patients subjected to OPT with gluten-containing croissants and those who underwent OPT with gluten-free croissants.
   J Clin Gastroenterol, 2016 · checked 2026-09-23
9. crs-r3-02 · randomised controlled trial, 32 healthy participants
   No further changes in any of the monitored variables or in the gut microbiome were detected.
   J Nutr, 2024 · checked 2026-09-23