Beginning with the well-known concept of a system at the third level of generalization, where it is associated with empirical engineering and science, several new system categories are proposed. These are related to deterministic and stochastic distinctions and show clearly whether system disturbances are internally or externally generated. They are therefore valuable in both input/output and structural approaches for developing conceptual system models and choosing mathematical formalisms. By combining these categories with the concepts of autonomy and non-autonomy, new and more precise definitions for stationarity and non-stationary are formulated. In this way it is shown how the steady-state properties of systems may be inferred from knowledge of their internal properties and from the steady-state levels of inputs or parameters. Several examples are presented to demonstrate how the proposed definitions usefully relate general system concepts to specific physical system properties.