Community violence is associated with health-compromising behaviors (HCBs) and adverse general health and pregnancy outcomes. It is unknown whether these effects persist after adjusting for health behaviors. Retrospective cohort study of 36,637 pregnancies, 2005-2009, using birth certificate/hospital discharge databases. Community violence rate = (nonaccidental injuries/total injuries) × 100%. ZIP codes were categorized into levels (CVL) by quartiles (lowest, 1, to highest, 4). Outcomes included HCBs (tobacco, alcohol, or drugs), fetal death, preterm birth, and infant small-for-gestational-age (SGA). Covariates included HCBs and maternal characteristics. The CVL median (interquartile range) was 5.5% (3.8%-8.8%). As CVL increased, rates of HCBs (8% vs 13% vs 14% vs 16%), fetal death (4 vs 5 vs 6 vs 8/1000 pregnancies), preterm birth (8% vs 9% vs 11% vs 13%), and infant SGA (8% vs 10% vs 13% vs 16%) increased. The odds of preterm birth (CVL1 = 1.00 [reference] vs CVL2 = 1.00 [0.88-1.14] vs CVL3 = 1.10 [0.96-1.26] vs CVL4 = 1.25 [1.09-1.42]) and infant SGA (CVL2 = 1.03 [0.93-1.17] vs CVL3 = 1.15 [1.01-1.30] vs CVL4 = 1.21 [1.07-1.38]) increased, after controlling for HCB. CVL is associated with fetal death, preterm birth, and infant SGA, independent of HCB. These findings may support the role of violence-reduction programs in improving pregnancy health behaviors and outcomes.