I provide evidence of the severe social costs imposed by infrastructure projects that are being implemented (i.e., projects started but not yet completed) in the context of sewerage in Peru. Using a counterfactual implementation predicted from geography-based cost considerations as an instrument, I show that implemented projects increase infant and under-five mortality. These results are driven by hazards, poor hygienic conditions and unsafe behavior, which increase deaths by waterborne diseases and accidents. Delays and mid-construction halting are common, and exacerbate the lethal effects of projects. Failing to take the implementation phase into account could severely bias the welfare evaluation of infrastructure.