Problems associated with the simulation of density fluctuations of limited breadth in a small cell are exposed and studied. The fluctuations are viewed as "physical clusters" of the type that might appear in nucleation processes and related phenomena. One of the most important features of the study stems from the fact that the simulation of a small heterogeneity in a macroscopic system presents problems that do not occur in the simulation of a bulk homogeneous property of the system. For example, once having simulated the probability of appearance of the fluctuation in a small cell, how is that result to be "mapped" onto the macrosystem in order to specify the equilibrium number of such fluctuations in that system? This problem is closely associated with the proper separation of the translational and internal degrees of freedom of the system, and has arisen in a number of fields, including the theory of nucleation. There are other problems associated with exponential dependence of cluster probability on the work of formation of the cluster, and also with rareness of some important clusters. In the latter case, simulative "umbrella sampling" does not always solve the entire problem. The present study is confined to clusters that appear in rarefied gases. Such systems are important in a number of scenarios, including nucleation processes. Several cluster models are considered including those consisting of molecules confined to a "container" of fixed volume and those constructed on the center of mass of the cluster. Connections between them are derived and rigorous solutions to the mapping problem are derived. Quantitative measures for the accuracy of approximate solutions, applied to cases in which the cluster is compact, are provided and exact solutions are provided even for the noncompact case. Some surprising results emerge from the study, among which is the fact that a cluster whose location is determined by one of its molecules, does not always have a probability of location that is uniform throughout a volume even if the cluster molecules do not see the boundaries of that volume. Also, the translational degrees of freedom of such a cluster cannot always be fully "activated" by freeing the "locational" molecule to move throughout the system.