Using case-control methodology, this study compared the characteristics of women who requested medically-trained birth attendants at home with those who did not, in order to identify constraints to service delivery and suggest program changes to increase service utilization in a rural area of Bangladesh. By several different measures, prior contact with medical professionals were much more common among cases than controls, with the greatest differences observed in the frequency of antenatal visits. At least one antenatal visit was the strongest predictor that a woman would call a medically-trained birth attendant, though primiparity, proximity to the provider, previous contact with the clinic, and education (both of the mother and her spouse) were also predictive. Recommendations include increasing contact through antenatal visits, and extending midwife coverage through training and supervision of traditional birth attendants. This study demonstrates the usefulness of case-control methodology in evaluating health care utilization.