Background Considering the changes in thyroid physiology associated with pregnancy and poor outcomes related to abnormal maternal thyroid function, international guidelines recommend using population-based trimester-specific reference intervals (RIs) for thyroid testing. If these RIs are not available in the laboratory, implementing recommended fixed cut-off values globally is still controversial. To address this issue, we aimed to establish appropriate RI of thyroid-stimulating hormone (TSH) in pregnant Turkish women for our laboratory and compare the prevalence of thyroid dysfunction based on the established and recommended criteria. Methods Of 2638 pregnant women, 1777 women followed in the obstetric outpatient were enrolled in the reference interval study after applying exclusion criteria related to medical and prenatal history. A retrospective study was conducted by collecting data from July 2016 to March 2019. Serum TSH was measured by UniCel DxI 800 Immunoassay System (Beckman Coulter Inc., Brea, CA, USA). The study design relied on two approaches in order to classify pregnant women: trimester-specific and subgroup-specific; the latter involved dividing each trimester into two subgroups: T1a, T1b, T2a, T2b, T3a, T3b. The lower and upper limits of the RIs were derived by the parametric method after normalizing the data distribution using the modified Box-Cox power transformation method. Results The lowest TSH value was detected at 8-12 weeks in early pregnancy, and the median value of TSH in the T1b subgroup was significantly lower than the T1a subgroup (P < 0.05). TSH levels showed a gradual trend of increase along with the pregnancy and increased significantly in the T2a, T2b, and T3b subgroups compared to the preceding subgroups (P < 0.05). Compared to the diagnostic criteria recommended by American Thyroid Association (ATA), the prevalence of thyroid dysfunction was significantly different from the established trimester- and subgroup-specific RIs throughout the pregnancy (P < 0.001). Conclusions We conclude that establishing gestation- and laboratory-specific RIs, especially for TSH, is essential for diagnosing thyroid disorders in pregnancy, and the recommended universal cut-off values, which may contribute to the risk of a misdiagnosis or a missed diagnosis, should be taken with caution in the clinical setting. However, regarding the fluctuation of thyroid function tests throughout pregnancy, trimester-specific RIs are insufficient, and implementing split phases is required.