We describe the study design and patterns of participation for a cohort study of preterm delivery, focused on genital tract infections, nutrition, tobacco use, illicit drugs and psychosocial stress. Women are recruited at 24–29 weeks’ gestation from prenatal clinics at a teaching hospital and a county health department. We recruited 57% of the first 1843 eligible women; 29% refused and 8% could not be contacted. White women were somewhat more likely to participate than African-American women (61% vs. 54% respectively). More notable differences were found comparing teaching hospital and health department clinics (71% vs. 47% participation respectively), with the health department clinic having a greater proportion refuse (24% vs. 33%) and more women who could not be contacted (4% vs. 11%). Participation was affected only minimally by day or timing of recruitment, but inability to contact diminished substantially as the study continued (13–0%). Refusals were largely unrelated to patient attributes. Lower education predicted inability to contact. Risk of preterm delivery was 14% among recruited women, 10% among women who refused, and 15% among women whom we were not able to contact, demonstrating that, overall, risk status was not lower among recruited women.