Abstract Objective. To test the hypothesis that the decision to use epidural analgesia during labor is influenced not only by the woman and her background but also by the local cultural practice in the delivery unit. Design. Population-based cohort study. Setting. All delivery units in Sweden. Population. A nationwide sample of 2,529 women. Methods. Data were collected by questionnaires in early pregnancy and two months after birth, and from the Swedish Medical Birth Register. Logistic regression analysis was conducted, adjusted for gestational age, induction of labor and infant birthweight. Main outcome measures. Epidural analgesia during labor. Results. The odds of having an epidural analgesia were more than twice as high in the Stockholm region (odds ratio (OR) 2.4; 95% confidence interval (CI) 1.7-3.4) and three times higher in middle-north Sweden (OR 3.0; 95% CI 1.7-5.3) compared with the south of Sweden. Of the maternal factors, nulliparity was the strongest predictor (OR 6.3; 95% CI 5.1-7.9), followed by a prenatal belief that epidural analgesia would be needed (OR 3.5; 95% CI 2.8-4.4). Conclusion. The hypothesis of the study was confirmed. The woman and her background as well as the local cultural practice in the delivery unit matter with regard to the use of epidural analgesia.