The most studied saffron extract: what that actually rests on
When a review concludes that saffron works, it has almost always pooled a dozen different saffron preparations made by different companies in different ways. That tells you something about a plant. It tells you very little about the specific material in the product in front of you.
What makes one saffron extract different is that the trials were run on it, rather than on saffron in general. Eleven randomised, placebo-controlled human trials have tested a single standardised extract, affron, in more than 1,000 people enrolled, aged 12 to 77.
Why one extract matters more than a plant
Two saffron extracts can be made from the same flowers and still be different ingredients. Solvent, temperature, plant part, drying and the compounds they are standardised to all change what ends up in the powder.
So a pooled analysis of twenty saffron studies is a finding about a genus. A trial record built on one standardised, per-batch-assayed extract is a finding about a material you can actually buy, and it is the only kind that transfers.
What the record actually consists of
Eleven placebo-controlled trials of the extract given on its own. Nine were run in Australia and two in Japan. The extract itself is Spanish, but no efficacy trial has been run in Spain.
The populations vary widely: teenagers with mild symptoms, adults reporting low mood, healthy recreationally active adults, adults dissatisfied with their sleep, and women aged 50 to 70. Trials ran from four to twelve weeks.
Two of the eleven are worth knowing about specifically, because they are the reason published counts of this literature disagree with each other. They are indexed in Japanese Pharmacology and Therapeutics rather than in the databases most people search, so a straightforward PubMed search returns nine and misses two. If you are checking a claim about how many trials exist, that is where the discrepancy comes from.
The two largest trials
The largest is a 2025 trial in 202 adults with subclinical depressive symptoms, run over twelve weeks at 28 mg a day. On the depression scale of the DASS-21, 72.3 per cent of the saffron group reached a reduction of at least seven points, against 54.3 per cent on placebo (p = 0.010). The effect size was modest at d = 0.39, and the authors noted a large placebo response. (Journal of Nutrition, 2025, PMID 40414301.)
The second is a 2026 trial in 86 women aged 50 to 70, also over twelve weeks and also at 28 mg a day. On the same scale the adjusted mean difference was 2.37 (95 per cent confidence interval 0.23 to 4.51), d = 0.48, p = 0.030, with 48.8 per cent reaching the seven-point reduction against 25.6 per cent on placebo (p = 0.026). (Lopresti and Smith, Frontiers in Nutrition, 2026;13:1838513.)
The one trial in young people
One trial in this set enrolled participants under 18: 80 young people aged 12 to 16 with mild to moderate symptoms and no diagnosis, of whom 68 completed. They took 14 mg twice a day, or placebo, for eight weeks.
On the young people's own ratings of the Revised Child Anxiety and Depression Scale, 37 per cent had their score at least halve, against 11 per cent on placebo (p = 0.014). That figure is the adolescents' self-report; their parents completed a separate version of the same scale, and the parent ratings did not consistently track the young people's own. (Lopresti et al., Journal of Affective Disorders, 2018, PMID 29510352.)
Two things to be straight about
The first is age. The youngest person enrolled anywhere in this literature is 12. There is no placebo-controlled trial of this extract in children under 12, and any claim that reads as though there is one is wrong.
The second is measurement. Mood in these trials is assessed with validated questionnaires rather than a blood test, which is standard practice in the field and is how essentially all mood research is done. The scales used here, the DASS-21 and the Revised Child Anxiety and Depression Scale, are the same validated instruments the field uses everywhere else.
What to take from it
Depth on one material beats breadth across a plant. Eleven placebo-controlled trials, more than 1,000 people, two countries, four to twelve weeks, ages 12 to 77, all on an extract with a published standardisation and per-batch assays.
When you next read that an ingredient is backed by research, the question to ask is whether the research was done on that ingredient, or on something with the same name.
Studies conducted on individual ingredients, not on any finished product. Individual results may vary.