BACKGROUND: Millions of people worldwide are drinking water with elevated arsenic concentrations. Epidemiologic studies, mainly cross-sectional in design, have suggested that arsenic in drinking water may affect pregnancy outcome and infant health. We assessed the association of arsenic exposure with adverse pregnancy outcomes and infant mortality in a prospective cohort study of pregnant women. METHODS: A population-based, prospective cohort study of 2924 pregnant women was carried out during 2002-2004 in Matlab, Bangladesh. Spontaneous abortion was evaluated in relation to urinary arsenic concentrations at gestational week 8. Stillbirth and infant mortality were evaluated in relation to the average of urinary arsenic concentrations measured at gestational weeks 8 and 30. RESULTS: : The odds ratio of spontaneous abortion was 1.4 (95% confidence interval [CI] = 0.96-2.2) among women with urine arsenic concentrations in the fifth quintile (249-1253 μg/L; median = 382 μg/L), compared with women in the first quintile (<33 μg/L). There was no clear evidence of increased rates of stillbirth. The rate of infant mortality increased with increasing arsenic exposure: the hazard ratio was 5.0 (95% CI = 1.4-18) in the fifth quintile of maternal urinary arsenic concentrations (268-2019 μg/L; median = 390 μg/L), compared with the first quintile (<38 μg/L). CONCLUSIONS: We found evidence of increased risk of infant mortality with increasing arsenic exposure during pregnancy, with less evidence of associations with spontaneous abortion or stillbirth risk.