Infrastructure for production, harvest, storage, and transportation of lignocellulosic biomass (LCB) is not developed. While some farmers have forage harvest machines and equipment that might be used to harvest LCB, it is unlikely that most regions would have machine harvest capacity necessary to provide massive quantities of LCB in a consistent package and to provide an orderly flow of LCB to a biorefinery throughout the year. Conventional models assume a fixed harvest charge per unit of LCB, and do not recognize that harvest machines may not be readily available and that harvest days are limited by weather. The objective of this research is to determine how the method of modeling harvest cost changes the estimate of the number of LCB harvest machines necessary to support a biorefinery and the estimated cost to deliver LCB. A mathematical programming model is developed that includes integer decision variables enabling investment in harvest machines that provide monthly harvest capacity based on expected harvest days per month. Results from a conventional model are compared to those of the alternative model that endogenously determines the number of harvest machines. To provide 3.628 dry kt daily to a biorefinery, the endogenous harvest cost model selected 26 harvest units. Alternatively, the solution provided by the conventional model that assumes a fixed cost per unit harvested and ignores weather constraints and machine investment provides a solution that in some months would require 55 harvest units and costs substantially more than the exogenously assumed fixed cost per unit. (C) 2008 Elsevier Ltd. All rights reserved.