This paper presents an application of genetic algorithms (GA) for solving the long-term power generation expansion planning (PGEP) problem, a highly constrained nonlinear discrete optimization problem. The problem is formulated into a mixed integer nonlinear programming (MINLP) program that determines the most economical investment plan for additional thermal power generating units over a planning horizon, subject to the requirements of power demands, power capacities, loss of load probability (LOLP) levels, locations, and environmental limitations. Computational results show that the GA-based heuristic method can solve the PGEP problem effectively and more efficiently at a significant saving in runtime, when compared with a commercial optimization package. Copyright (c) 2005 John Wiley & Sons, Ltd.