This paper examines the effect of water-borne lead exposure on infant mortality in American cities over the period 1900-1920. Infants are highly sensitive to lead, and more broadly are a marker for current environmental conditions. The effects of lead on infant mortality are identified by variation across cities in water acidity and the types of service pipes that the water ran through – lead, iron, or concrete – which together determined the extent of lead exposure. Estimates that restrict the sample to cities with lead pipes and panel estimates provide further support for the causal link between water-borne lead and infant mortality. The magnitudes of the effects were large. In 1900, a decline in exposure equivalent to an increase in pH from 6.675 (25th percentile) to 7.3 (50th percentile) in cities with lead-only pipes would have been associated with a decrease in infant mortality of 7 to 33 percent or at least 12 fewer infant deaths per 1,000 live births. This paper examines the effect of water-borne lead exposure on infant mortality in American cities over the period 1900-1920. Infants are highly sensitive to lead, and more broadly are a marker for current environmental conditions. The effects of lead on infant mortality are identified by variation across cities in water acidity and the types of service pipes that the water ran through – lead, iron, or concrete – which together determined the extent of lead exposure. Estimates that restrict the sample to cities with lead pipes and panel estimates provide further support for the causal link between water-borne lead and infant mortality. The magnitudes of the effects were large. In 1900, a decline in exposure equivalent to an increase in pH from 6.675 (25th percentile) to 7.3 (50th percentile) in cities with lead-only pipes would have been associated with a decrease in infant mortality of 7 to 33 percent or at least 12 fewer infant deaths per 1,000 live births.