Context: Although thyroid hormone replacement may improve outcomes in pregnant women with subclinical hypothyroidism (SCH), the extent to which they receive treatment is unknown. Objective: To describe levothyroxine (LT4) treatment practices for pregnant women with SCH. Design: Retrospective cohort study. Setting: Large US administrative claims database. Participants: Pregnant women with SCH defined by untreated TSH 2.5 to 10 mIU/L. Main Outcome Measure: Initiation of LT4 as a function of treating clinician specialty (endocrinology, obstetrics/gynecology, primary care, or other), baseline TSH, patient clinical and demographic factors, and US region. Results: We identified 7990 pregnant women with SCH; only 1214 (15.2%) received LT4. Treatment was more likely in patients with higher TSH, obesity, recurrent pregnancy loss, thyroid disease, and cared for by endocrinologists. Proportion of treated women increased over time; LT4 treatment was twice as likely in 2014 as in 2010. Women in Northeast and West US were more likely to receive LT4 compared with other regions. Asian women were more likely, whereas Hispanic women were less likely, to receive LT4 compared with white women. Endocrinologists started LT4 at lower TSH thresholds than other specialties, and treated women who were more likely to have had recurrent pregnancy loss and thyroid disease than women treated by other clinicians. Conclusions: We found large variation in the prescription of LT4 to pregnant women with SCH, although most treatment-eligible women remained untreated. Therapy initiation is associated with geographic, clinician, and patient characteristics. This evidence can inform quality improvement efforts to optimize care for pregnant women with SCH.