The authors evaluated the association of prenatal arsenic exposure with size at birth (birth weight, birth length, head and chest circumferences). This prospective cohort study, based on 1,578 mother-infant pairs, was conducted in Matlab, Bangladesh, in 2002-2003. Arsenic exposure was assessed by analysis of arsenic in urine collected at around gestational weeks 8 and 30. The association of arsenic exposure with size at birth was assessed by linear regression analyses. In analysis over the full range of exposure (6-978 microg/L), no dose-effect association was found with birth size. However, significant negative dose effects were found with birth weight and head and chest circumferences at a low level of arsenic exposure (<100 microg/L in urine). In this range of exposure, birth weight decreased by 1.68 (standard error (SE), 0.62) g for each 1-microg/L increase of arsenic in urine. For head and chest circumferences, the corresponding reductions were 0.05 (SE, 0.03) mm and 0.14 (SE, 0.03) mm per 1 microg/L, respectively. No further negative effects were shown at higher levels of arsenic exposure. The indicated negative effect on birth size at a low level of arsenic exposure warrants further investigation.