This study looked at women with a thin uterine lining undergoing frozen embryo transfers and compared three ways to prepare the lining: natural cycle, artificial hormone cycle, and an artificial cycle with a GnRH‑agonist (similar to gonadorelin). Overall pregnancy and birth rates were about the same, but the natural cycle had fewer early pregnancy losses than the artificial cycle. For the thinnest linings (<7 mm), none of the methods showed any advantage.

Endometrial thickness independently predicts pregnancy outcomes in frozen embryo transfer (FET) cycles. Thin endometrium always results in implantation failure and worse obstetric-neonatal outcomes. However, it has not been reported which endometrial preparation strategy achieved optimal outcomes in patients with thin endometrium undergoing FET cycles. This historical cohort study was conducted on 2671 women with thin endometrium who underwent their first FET cycle at the Reproductive Medicine Center of a university-affiliated hospital between January 2018 and August 2022 (followed up to August 2023). Patients were divided into three groups according to endometrial preparation protocols (NC: natural cycle, AC: artificial cycle, GnRH-a&#x2009;+&#x2009;AC: AC with gonadotropin-releasing hormone agonist pretreatment). Thin endometrium was defined as endometrial thickness&#x2009;&lt;&#x2009;8&#xa0;mm on the first day of progesterone administration. Patients with uterine abnormalities, recurrent spontaneous abortion, or donor oocytes were excluded. We also further analyzed the condition of endometrial thickness&#x2009;&lt;&#x2009;7&#xa0;mm. Pregnancy and obstetric-neonatal outcomes were assessed. A total of 2671 patients were included in the study. Among patients with endometrial thickness &lt; 8 mm, the clinical pregnancy rate was 36.2% (691/1908) in the AC group, 35.2% (178/506) in the GnRH-a + AC group, and 33.9% (87/257) in the NC group. The live birth rates were 26.8% (512/1908), 25.3% (128/506), and 27.6% (71/257) in the three groups, respectively. No statistical differences were observed in pregnancy rates or obstetric-neonatal outcomes in pairwise comparisons, except that the biochemical pregnancy loss rate in the NC group was significantly lower than that in&#xa0;the&#xa0;AC group (3.9% versus 8.6%, <i>P</i> &lt; 0.05). Furthermore, this result remained consistent after multivariate logistic regression (crude odds ratio [95% CI]: 0.428 [0.223,0.821], adjusted odds ratio [95% CI]: 0.444 [0.230,0.856]). For patients with endometrial thickness &lt; 7 mm, there were no significant differences in any outcomes across the&#xa0;three groups. Analysis using the 8 mm cut-off revealed a lower&#xa0;biochemical pregnancy loss rate in the NC group compared to the AC group.&#xa0;In contrast,&#xa0;no significant differences were observed in clinical pregnancy, live birth, or obstetric-neonatal outcomes based on endometrial preparation strategy for patients with an endometrial thickness &lt; 7 mm or 8 mm. The online version contains supplementary material available at 10.1186/s12978-025-02166-z.