What the studies show

These interview studies of people who stopped or reversed a gender transition capture what surveys and registries cannot: reasons, shifting identities and experiences of care. Across three samples, several themes recur. Body discomfort could return after stopping, or turn into distress about permanent changes such as a deeper voice (Pullen Sansfaçon et al., 2023). They describe same-sex attraction, sometimes with shame about being gay, trauma, autism or mental illness as factors that fed their distress but were not explored by clinicians (Pullen Sansfaçon et al., 2023; MacKinnon et al., 2023). They report feeling under-informed about risks before treatment (Lomax & Butler, 2025; MacKinnon et al., 2023). Many stopped hormones without medical supervision and turned to online detransition communities for support (MacKinnon et al., 2022; Lomax & Butler, 2025). Identities after stopping were varied. In the Canadian sample, 60% moved from a binary trans identity to a nonbinary one (MacKinnon et al., 2023).

Accounts of regret differ by sample. In the Canadian study, 6 of 27 participants who medically transitioned (22%) were judged to regret it and 3 (11%) had mixed feelings, while 18 (67%) expressed no regret (MacKinnon et al., 2022). That sample was recruited broadly, and 75% still identified as trans. The UK study was smaller and recruited through detransition networks. Its six women, four of whom had had double mastectomies, said hormones and surgery had not resolved their distress or changed their sex. Some also came to see narrow ideas of womanhood as having shaped their original decision (Lomax & Butler, 2025).

Limits of the evidence

The evidence base is very small: three independent samples totaling 54 people, because the two Canadian papers re-analyze the same 28 interviews. All samples were self-selected, mostly through social media and detransition or trans networks, and were predominantly born female. Accounts are retrospective, given at a single interview and not checked against medical records. "Regret" was coded by researchers from conversation rather than measured with a validated instrument, so none of these figures are rates. Several teams acknowledge that their own positions may have shaped the accounts and themes. These studies generate hypotheses and describe unmet needs. They cannot show how common any experience is.