We examine issues involved in applying and interpreting free-energy perturbation (FEP) calculations in molecular simulation. We focus in particular on the accuracy of these calculations, and how the accuracy differs when the FEP is performed in one or the other direction between two systems. We argue that the commonly applied heuristic, indicating a simple average of results taken for the two directions, is poorly conceived. Instead, we argue that the best way to proceed is to conduct the FEP calculation in one direction, namely that in which the entropy of the target is less than the entropy of the reference. We analyze the behavior of FEP calculations in terms of the perturbation-energy distribution functions, and present several routes to characterize the calculations in terms of these distributions. We also provide prescriptions for the selection of an appropriate multistage FEP scheme based on how the important phase-space regions of the target and reference systems overlap one another.