OBJECTIVE (1) To determine the prevalence of antidepressant utilisation before, during, and after pregnancy, (2) to determine switches, dosages, and classes of antidepressant used during pregnancy, and (3) to identify factors associated with their use at the beginning and at the end of pregnancy. METHODS Retrospective longitudinal cohort. METHODS The 'Medication and Pregnancy' cohort was used for this study. This cohort was built by the linkage of three administrative databases (Régie de l'Assurance Maladie du Québec [RAMQ], Med-Echo, and l'Institut de la Statistique du Québec). METHODS All pregnancies occuring in Quebec between January 1 1998 and December 31 2002. METHODS Date of entry in the cohort was the first day of gestation. To be eligible for this study, women had to be (1) 15-45 years old at cohort entry and (2) covered by the RAMQ drug plan for at least 12 months before, during, and at least 12 months after pregnancy. Antidepressant users were defined as those receiving at least one antidepressant before, during, or after pregnancy, depending on the time period analysed. Logistic regression models were used to identify factors associated with receiving an antidepressant either at the beginning or at the end of pregnancy. METHODS To determine the prevalence and predictors associated with the use of antidepressants. RESULTS A total of 97,680 women met inclusion criteria. The prevalence rates significantly declined during the first trimester compared with before pregnancy (3.7 versus 6.6%, P < 0.01). During pregnancy, antidepressants were used under the recommended daily dosage 7.7% of the time, and 4.7% of women switched to another class of antidepressant. Factors significantly associated with antidepressant utilisation on the first day of gestation (P < 0.05) were older maternal age, being on welfare, and calendar year; receiving at least six different types of medications other than antidepressants, having at least two different prescribers, having at least three visits to the physician, and having at least one diagnosis of depression in the year before pregnancy also increased the odds of having an antidepressant. Similar predictors were found at the end of pregnancy. CONCLUSIONS Our findings indicate that antidepressant utilisation declines once pregnancy is diagnosed.