This article develops a new, general account of replication (the Resampling Account of replication). I argue that a replication is an experiment that resamples the experimental components of an original experiment that are treated as random factors and that the function of replications is, narrowly, to assess the reliability of the replicated experiments. On this basis, I argue that the common notion of conceptual replication is confused and that the ongoing controversy about the relative value of direct and conceptual replications should be dissolved.