A new technique is presented for probabilistically simulating the operation of power systems which include multiple assigned-energy units (e. g., conventional hydroelectric plants or groups of plants, and, under certain circumstances, nuclear units, conventional thermal units operating under " take-or-pay" fuel supply contracts, and cogeneration units). Within the context of the probabilistic simulation, the new technique is completely rigorous. Statistical cumulants are used to represent equivalent load. The new technique has been successfully applied in an advanced technology evaluation and expansion model--the TVA TARANTULA model. This application has been accomplished without sacrificing other simulation capabilities such as random partial outages, maintenance scheduling based on LOLP and short-term load forecast uncertainty, and the ability to aggregate (without loss of rigor) groups of many small units for purposes of computational expediency. A description is provided of tests performed to evaluate the new technique.