A general probabilistic model of a two-site wind-energy conversion system (WECS) is described in this paper. The wind speeds at the two sites are not assumed to be independent, which prevents the convolution theorem from being directly applicable. Instead, a model based on the conditional probability theorem and mutually exclusive events is presented. The model supports the assessment of the energy resource available to supply a given load represented by its load duration curve, taking into consideration wind-turbine failure modes and the intermittent nature of the wind resource. The model is based on a capacity-probability table constructed using fundamental probability theorems on conditional and mutually exclusive events. The table is then combined with a load model in order to assess its performance and determine the expected energy not supplied (EENS) of the system. A planning study is then reported on a realistic system to illustrate how the most efficient wind turbines can be selected to support an existing thermal system using a procedure based on a minimization of the incremental marginal cost of operation.