Objectives. Elevated arsenic levels in tube-well water in Bangladesh have prompted extensive mitigation projects. We evaluated the effectiveness of long-term mitigation efforts by longitudinally measuring arsenic exposure in pregnant women and their children, the most susceptible population groups. Methods. The study was nested in a population-based nutrition intervention in Matlab, Bangladesh. Mother–child pairs (n = 1951) were followed from 2001 to 2003, beginning in early gestation and continuing to 5 years postpartum. We measured arsenic concentrations in urine (U-As) of the 5-year-old children by using high-performance liquid chromatography online with hydride generation and inductively coupled plasma mass spectrometry and compared them with earlier childhood U-As and maternal U-As during pregnancy. Results. Children had elevated U-As at 5 years old (median = 51 μg/L, 5th–95th percentiles = 16–355 μg/L), and U-As distribution was similar to that observed in the mothers during gestation. Children's U-As at 5 years old significantly correlated with their U-As at 1.5 years old and to maternal U-As during early and late gestation. Conclusions. Despite major mitigation efforts, arsenic exposure remains highly elevated in rural Bangladesh. Further mitigation strategies are required and must be rigorously evaluated for long-term efficacy.