To assess epidemiologic risk factors for preterm birth subcategories in an urban population, we undertook a study of 31,107 singleton livebirths that took place at Mount Sinai Hospital in New York City between 1986 and 1994. We subdivided the preterm births into preterm premature rupture of the membranes, preterm labor, and medically induced births. We obtained information regarding the preterm subtypes and their epidemiologic risk factors from a computerized perinatal database. Adjusted odds ratios showed an increased risk for all three preterm birth subtypes in women who were black (1.9 for preterm premature rupture of membranes, 2.1 for preterm labor, and 1.7 for medically induced births) or Hispanic (1.7 for preterm premature rupture of membranes, 1.9 for preterm labor, and 1.6 for medically induced births), those who had had a previous preterm birth (3.2 for preterm premature rupture of membranes, 4.5 for preterm labor, and 3.3 for medically induced births), those who began prenatal care after the first trimester ( 1.4 for preterm premature rupture of membranes, 1.3 for preterm labor, and 1.3 for medically induced births), women who had been exposed to diethylstilbestrol in utero (3.1 for preterm premature rupture of membranes, 4.1 for preterm labor, and 3.7 for medically induced births), patients with preexisting diabetes mellitus (2.2 for preterm premature rupture of membranes, 2.4 for preterm labor, and 9.5 for medically induced births), and those with antepartum bleeding (2.8 for preterm premature rupture of membranes, 3.6 for preterm labor, and 3.7 for medically induced births). Other sociodemographic, constitutional, life-style, and obstetrical characteristics differed across the groups. Variation in some of the risk factors among the preterm subtypes implies that epidemiologic assessment of the more specific outcomes would be advisable.