Background: Multiple sclerosis (MS) is frequently diagnosed during women's reproductive years, raising the question of safety of Disease Modifying Therapies (DMTs) during pregnancy. To assess the use and safety of DMTs within population databases, women with MS must be identified. Objectives: To explore strategies for identifying MS patients and calculating MS prevalence in large-population databases. Methods: This pilot study was conducted using data from EFEMERIS, a French cohort of pregnant women, between 2005 and 2019. Five algorithms to identify MS and 3 methods to calculate MS prevalence were tested. The algorithm MS1 required at least one prescription for a MS-specific DMT or at least one MS diagnosis, and the number of components required increased from MS1 to MS3, MS4 required at least 1 diagnosis, and MS5 required at least 2 diagnoses. Methods used for prevalence calculation were: period prevalence (number of cases/total number of women), person-time prevalence (number of person-days after MS identification /total number of person-days), and average prevalence (average of point prevalence calculated on each first day of the month). Results: The study included 103,332 women. 67 women had at least one MS diagnosis, and 63 women had at least one prescription for a MS-specific DMT. The number of women diagnosed with MS according to the different algorithms were: 105 for MS1, 75 for MS2, 57 for MS3, 67 for MS4 and 19 for MS5. Using MS1 to identify MS cases, the median follow-up of women with MS was 360 days and MS prevalence per 100,000 women (95% confidence interval) according to the different methods was: 101.6 (101.4–101.8) for the period prevalence, 53.9 (53.94–53.96) for the person-time prevalence, and 49.0 (47.7–50.2) for the average prevalence. Conclusions: MS prevalence in EFEMERIS was lower than previously published for France (in a study by Foulon and colleagues, prevalence was 210.0 per 100,000 women in the general population). This was expected because women included in EFEMERIS have short follow-up, and, consequently, a short time frame in which the disease might be identified. Also, in our study only women who are pregnant are included, and MS may be less common in this population than in the general female population. The prevalence was lower when time was considered (person-time and average methods) compared with the period prevalence, which may be explained by the MS identification being predominantly at the end of follow-up. The choice of the algorithm for identifying MS and the method for prevalence calculation must be considered and adapted to each database. This study will be replicated in 7 other large databases in the coming months.