Most modern computer games provide a virtual environment as a context for player interaction. Recently, many multi-player online games have adopted the persistent-state gaming model, which provides a central virtual environment with essentially infinite lifetime. However, a displeasing part of these long-lasting environments is that, like their predecessors, they are still assumed to be static, unchanging even in the long-term. In response to this fact, we introduce the adaptive virtual environment which automatically adapts based on activity occurring within the environment. In computer games, adaptive virtual environments are systems that correspond to real-world physical or social systems. These systems are computationally formalized by adhering to a generic adaptation model containing abstract components and procedures. Herein, as a proof of concept, we design and analyze the behavior of two adaptive versions of such systems commonly found in persistent-state games. To achieve this, we build an implementation of an abstract interactive simulator that applies the adaptation process to our example systems. Each system is internally represented as a plug-in module containing system-specific implementations of the model's abstractly-defined procedures. Performance of the adaptation process is then evaluated using simulation data. Finally, improvements such as optimizations and better movement models for agent simulation are investigated, and the general usefulness and applicability of the concepts is discussed.