The precipitation efficiency of cumulus congestus clouds is investigated with a new Lagrangian particle framework for large-eddy simulations. The framework is designed to track particles representative of individual water molecules. A Monte Carlo approach facilitates the transition of particles between the different water classes (e.g., vapor, rain, or graupel). With this framework, it is possible to reconstruct the pathways of water as it moves from vapor at a particular altitude to rain at the surface. By tracking water molecules through both physical and microphysical space, the precipitation efficiency can be studied in detail as a function of height. Large-eddy simulations of individual cumulus congestus clouds show that the clouds convert entrained vapor to surface precipitation with an efficiency of around 10%. About two-thirds of all vapor that enters the cloud does so by entrainment in the free troposphere, but free-tropospheric vapor accounts for only one-third to one-half of the surface rainfall, with the remaining surface rainfall originating as vapor entrained through the cloud base. The smaller efficiency with which that laterally entrained water is converted into surface precipitation results from the smaller efficiencies with which it condenses, forms precipitating hydrometeors, and reaches the surface.