In mild to moderate depression, more than one review found that a saffron extract did more than a placebo. In several trials set beside an SSRI, the easing of symptoms looked similar. Those trials were small. Looking similar is not the same as knowing the two are equal, and the studies were not built to settle that.
Twelve studies
Ahead of a placebo.
A 2020 review brought together 12 studies of mild to moderate depression. People on saffron came out ahead of people on a placebo. Where the same review looked at synthetic antidepressants, its own summary called saffron as effective as those drugs, and side effects were not clearly different from placebo or from the medicines. The authors still asked for larger, longer trials in more than one center. That request is part of the kindness of the paper. A small good result is not a promise for everyone.
PubMed 32221179
Forty-six trials
The difference had a number, and a lighter score.
A 2024 review gathered 46 randomized trials, from four weeks to forty-eight, across several conditions. Against placebo, the depression effect size was −4.26, with a confidence interval from −5.76 to −2.77. On these forms a negative number means people felt lighter, not heavier. The review called saffron non-inferior to usual medicines. That is the authors’ word for “not worse, in the trials they pooled.” The face-to-face comparisons under that word are fewer, and smaller.
PubMed 38424688
Beck and Hamilton
What people said about themselves moved. The pooled clinician scale did not.
A 2026 review of 34 randomized trials and 1,769 adults found Beck depression scores lower across 14 trials and 817 people, a weighted mean difference of −4.39 (confidence interval −6.64 to −2.15). The Beck form is one a person fills in about their own days. In that same review the Hamilton scale, which a clinician scores, did not show a clear change once the trials were combined. Many single trials had used the Hamilton scale and reported lighter scores. Put together, that signal faded. The reviewers called the certainty moderate, and the trials did not all agree with each other. Both halves matter if you are trying to be honest with yourself.
The 2026 review · PubMed 41693488
Eight studies against an SSRI
No clear gap between the scores.
Short trials, often against fluoxetine or imipramine, looked broadly close over about six weeks. A 2026 review says those trials were small, underpowered, and not designed to prove saffron is as good as the medicine. A separate look at eight depression studies against SSRIs found a standardized difference of 0.10, with a confidence interval from −0.09 to 0.29. That range includes “no difference at all.” People on saffron reported fewer unwanted effects in that comparison, a risk difference of −0.06. Close, in a small trial, is not a reason to set a prescribed medicine down.
PubMed 41939549 · PubMed 38913392
What the trials used
About 28 to 30 milligrams a day.
About 30 milligrams a day of a prepared extract, for about six weeks, is the pattern a 2026 review keeps meeting. A larger 2025 trial used 28 milligrams a day for 12 weeks. Other saffron trials run anywhere from four weeks out to forty-eight, so six or twelve weeks is common, not a rule for every study. These are descriptions of what researchers gave. They are not an amount for you to take from this page.
PubMed 41939549 · PubMed 40414301
A larger, milder trial
Two hundred and two adults, and a large lift on placebo too.
In 2025, 202 adults whose low mood was real but not a full clinical depression took 28 milligrams a day of a standardized extract from the red threads, the one the paper calls Affron, or a placebo, for 12 weeks. On the DASS-21 depression form, scores fell further on the extract, by about 2.92 points (confidence interval −5.13 to −0.71). A drop of 7 points or more happened for 72.3 percent on the extract and 54.3 percent on placebo. The authors said that placebo lift was large, and that it has to be faced, not waved away. Hope itself did a lot of the work. A smaller 2025 trial, 51 healthy adults with a similar mild low mood, used about 30 milligrams a day for six weeks and did not see a clear change on its main mood score. A bigger trial is welcome. It does not mean every mild-mood study agreed.
PubMed 40414301 · PubMed 41047129