Objective: Contraception for women with epilepsy (WWE) is important, but challenging; anti-epileptic drugs (AED) and hormonal contraceptives interact. The levonorgestrel-releasing intrauterine device (LNG-IUD) has received little study in chronic illnesses such as epilepsy.  We explored LNG and serum hormone binding globulin (SHBG) levels, bleeding patterns and endometrial thickness in WWE using the LNG IUS and co-administered AEDs. Methods: We conducted this pilot study in 20 WWE at Columbia University Irving Medical Center in New York City. We included women ages 18 to 45 years with well-controlled epilepsy and a stable AED regimen initiating the LNG-IUD (20mcg/day) for contraception. Participants provided written informed consent prior to study procedures. We collected blood and measured the endometrium by transvaginal sonography before IUD placement and at 3 weeks, 12 weeks and 24 weeks thereafter. Participants recorded bleeding and spotting days in paper diaries. We categorized AED regimens as enzyme-inducing if carbamazepine, oxcarbazepine, or topiramate was included and others as non-inducers. The primary outcomes were bleeding patterns, endometrial thickness, serum LNG and sexual hormone binding globulin (SHBG) levels over time (descriptive analysis). We compared these primary outcomes between WWE using inducer (5 participants) and non-inducer AED regimens (n=15) using mixed-effects linear regression modeling. Serum LNG was measured by radioimmunoassay, SHBG by an immunoassay (DELFIA). Results: Twenty WWE completed follow-up visits and are included in this analysis. The median endometrial thickness, LNG and SBHG levels remained stable during 24 weeks of LNG-IUD use. Median SHBG levels were higher in WWE using inducer regimens than those using non-inducer regimens throught the study [baseline: 59.1 nmol/L (IQR: 52.8-68.5) vs. 39.1 nmol/L (IQR: 31.9-47.3), p=0.03]. LNG levels and endometrial thickness were comparable between WWE using inducer and non-inducer regimens. The number of bleeding and spotting days decreased during 6 months of LNG-IUD, regardless of AED inducer status.  Participants reported a median of 16 bleeding or spotting days (IQR 13-23) in the first 30-day reference period (RP), with 6.5 days (4-8.5) of bleeding-only. By the last 30 days (RP 6), the median number of bleeding and spotting days decreased to 7 (IQR 2.5-10), with 2 days (0.5-4) of bleeding-only. Conclusion: This pilot study suggests that WWE who initiate the LNG IUS experience stable LNG levels and gradually decreasing bleeding over the first six months of use. Regimens with an enzyme-inducing AED were not associated with a clinically significant difference in bleeding patterns, endometrial thickness, or serum LNG and SHBG levels.