A multi-product cost function model was used to analyze energy demand in various agricultural subsectors. This approach has advantages over previously used approaches since it reduces aggregation bias, considers technological jointness, and provides various disaggregative measures related to energy input demand. When fitted to West Virginia county level data, labor and miscellaneous inputs in crop and livestock production were found to be substitutes for energy, while capital, machinery, and fertilizer were complementary to energy. Energy demand was inelastic and increases in machinery prices had the largest reduction effect on energy demand. Technological change was found to be capital, machinery, and fertilizer using, but it was labor and energy saving. Analyses indicated that the elasticity of demand for energy inputs with respect to livestock output was significantly larger than the elasticity with respect to crop output.