What the studies show
Evidence that testosterone and estrogen improve mental health is consistently rated weak. A review of 20 studies found associations with better quality of life and less depression and anxiety, but graded every conclusion low strength and could reach none on suicide (Baker et al., 2021). A later review of 24 studies in people under 26 found no randomized trials, rated most mental-health findings very low certainty, and concluded that neither benefit nor harm could be ruled out (Miroshnychenko et al., 2025). Continuation figures vary sharply by setting: 98% of 720 Dutch adolescents who had first taken puberty blockers still held a prescription at a median age of about 19–20 (van der Loos et al., 2022), and a Boston clinic recorded 93% of 1,050 adolescents as continuing, although 210 of them had not been seen for over a year (Boskey et al., 2025). By contrast, about 30% of 952 US military-insured patients stopped within four years, testosterone users more often than estrogen users (Roberts et al., 2022). None measured regret or wellbeing.
Safety findings are more consistent. Trans women on estrogen had venous blood clots at 4.5 to 5.5 times expected population rates in an Amsterdam cohort (Nota et al., 2019), and a US cohort found the excess growing to about 17 extra clots per 1,000 compared with non-trans men after eight years (Getahun et al., 2018). Breast cancer in trans women was about 46 times the male rate but about a third of the female rate, based on 15 cases (de Blok et al., 2019), and the antiandrogen cyproterone acetate was linked to 7 to 19 times the risk of meningioma (Mikkelsen et al., 2022). In trans men, testosterone worsened cholesterol (Maraka et al., 2017), and the cumulative chance of exceeding the male red-cell limit reached 38% after ten years (Madsen et al., 2021).
Limits of the evidence
Almost all of it is observational: single-clinic cohorts, registries adjusted for few variables, and self-selected surveys. Follow-up is often short, loss to follow-up is high, and people who stopped hormones are frequently excluded from safety analyses. An Australian registry reporting four to nine times general-population mortality could not separate the effect of hormones from underlying illness or other factors (Saxby et al., 2026).