A linear programming-based approach was applied to optimize the power (or load) distribution among 200-MW and 310-MW units of a 1330-MW power plant (co-)firing fuel oil and natural gas. The objective function for the optimization was the total costs, including "internal" (or fuel) and "external" (or environmental) costs, for the power plant meeting the current load demand. Both fuel and environmental costs (the latter being associated with NOx, SO3, SO2 and CO2 emissions) of the units were predicted for different fuel options. The total costs for an individual boiler unit were shown to be quasi-linear with the unit load. Optimal load dispatch over the units was determined using a typical daily load curve of the power plant, for distinct climatic seasons in Thailand. The time-domain savings of the total costs resulted from switching the units from actual to the optimized loads were quantified for this power plant. (c) 2007 Elsevier B.V. All rights reserved.