Hair Relaxer and Cancer Evidence: What the Studies Found—and Their Limits

Start with study design, not the headline

Most of the prominent evidence comes from prospective cohorts. Participants reported product use before the later cancer diagnosis, which reduces the risk that a diagnosis changed how they remembered prior exposure. Researchers then compared cancer incidence across reported use patterns while adjusting for measured factors. That design can reveal an association, but people are not randomly assigned to use hair relaxers, and users may differ from nonusers in ways a model cannot fully capture.

Exposure measurement is a central limitation. 'Straighteners, relaxers, or pressing products' can combine products with different chemistry. Brand and ingredient lists change over time, salon services may mix products, and a baseline questionnaire cannot perfectly capture childhood use or changes over a decade. A study may support a population-level hypothesis while remaining unable to identify which chemical, dose, formulation, or exposure window matters.

Uterine cancer: two cohorts, related but not identical results

The 2022 Sister Study analysis included 33,947 participants with a uterus at enrollment and identified 378 uterine cancers over an average 10.9 years. Ever use in the prior year was associated with a hazard ratio of 1.80 compared with never use; frequent use, defined as more than four times in the prior year, had a hazard ratio of 2.55. The confidence intervals were wide, and the authors described their work as the first epidemiologic evidence on this question and called for replication.

The 2023 Black Women's Health Study followed 44,798 Black women and identified 347 uterine cancers. The overall estimate for heavy use was 1.18, with a confidence interval that included 1.0. Elevated associations appeared among postmenopausal women for moderate, heavy, and very long-term use, while premenopausal results were null. Differences in population, exposure definitions, follow-up, outcome mix, and statistical precision mean the two studies should be weighed together, not portrayed as exact duplicates.

Ovarian and breast cancer findings need their own lanes

A 2021 Sister Study analysis of 40,559 participants reported no ovarian-cancer association for ever use of the examined hair products. Frequent use of straighteners, relaxers, or pressing products was associated with higher ovarian-cancer incidence, with a hazard ratio of 2.19 and a confidence interval from 1.12 to 4.27. Only 241 ovarian cancers occurred, and fewer fell within the frequent-use comparison, which limits precision and makes replication particularly important.

A separate Sister Study paper reported associations between permanent dye or straightener use and breast cancer, with patterns that varied by race and product type. Breast cancer is not a stand-in for uterine or ovarian cancer: tumor biology, hormonal pathways, background incidence, and competing risk factors differ. A litigation page that cites a breast-cancer study to support every gynecologic claim—or the reverse—overstates what the research measured.

Newer endpoints widen the questions, not the certainty

A 2026 Sister Study analysis examined non-reproductive cancers among 46,287 participants. It reported associations with thyroid cancer, non-Hodgkin lymphoma, and pancreatic cancer, and little to no evidence for several other sites. Because the researchers tested multiple outcomes and some cancers were uncommon, the results need confirmation. They do not automatically expand the injuries managed in MDL 3060 or establish that the same exposure pathway explains every reported association.

A 2025 analysis focused on uterine fibroids among Black women in the Sister Study, including use at ages 10 to 13 and later incident diagnoses. Fibroids are noncancerous and cannot be used as a synonym for uterine cancer. Such work may help researchers investigate endocrine or inflammatory pathways and timing, but a biological hypothesis must still be tested, and a nonmalignant outcome does not prove a malignant one.

Turning epidemiology into an individual legal claim

Courts typically separate general causation—whether an exposure is capable of causing an outcome—from specific causation—whether it caused this person's disease. Epidemiology can inform the first question, but the second may involve product identification, formulation, frequency and duration, latency, pathology, age, obesity, reproductive and hormonal history, family history, genetics, and other exposures. Experts must use reliable methods and explain how population evidence fits the individual record.

Relative measures also need absolute context. A hazard ratio does not mean that the same percentage of users will develop cancer, and it cannot be multiplied by a case count to estimate future diagnoses. NIH translated the uterine finding into an illustrative cumulative-risk comparison under specified assumptions, but it still warned that more research was needed. People with health concerns should discuss symptoms and screening with a clinician; litigation research is not a diagnostic tool and should not prompt unsupervised medical decisions.

Reader questions

Frequently asked questions