The authors evaluated the effect of arsenic exposure on fetal and infant survival in a cohort of 29,134 pregnancies identified by the health and demographic surveillance system in Matlab, Bangladesh, in 1991-2000. Arsenic exposure, reflected by drinking water history and analysis of arsenic concentrations in tube-well water used by women during pregnancy, was assessed in a separate survey conducted in 2002-2003. Data on vital events, including pregnancy outcome and infant mortality, were collected by monthly surveillance at the household level. The risk of fetal loss and infant death in relation to arsenic exposure was estimated by a Cox proportional hazards model. Drinking tube-well water with more than 50 microg of arsenic per liter during pregnancy significantly increased the risks of fetal loss (relative risk = 1.14, 95% confidence interval: 1.04, 1.25) and infant death (relative risk = 1.17, 95% confidence interval: 1.03, 1.32). There was a significant dose response of arsenic exposure to risk of infant death (p = 0.02). Women of reproductive age should urgently be prioritized for mitigation activities where drinking water is contaminated by arsenic.