Korea imports all of its primary energy, which leads to high dependency and vulnerability related to its energy supply. Efficiency in the use of energy is a way to reduce dependency and emissions. This study provides empirical results of the stochastic production process in energy use. Special attention is given to the factors that increase the risk or variation of using more of the energy input in production. A dynamic panel model is specified and applied to 25 Korean industrial sectors over the period 1970-2007. The determinants of energy use are identified and their effects in the form of elasticities of energy use are estimated. Stochastic production technology is applied to estimate an energy demand model based on an inverted factor demand. The findings reveal that: first, there are large variations in the degree of overuse or inefficiency in energy use among the individual industries as well as over time; second, information and communication technology (ICT) capital and labor are substituting for energy; and third, ICT capital input decreases the variability of energy demand while non-ICT capital, material and labor increase the variability of energy demand. The results suggest that technical progress contributes more to the increase in the mean energy demand than to the reduction in the level of risk. It is recommended that industries increase their level of ICT capital as well as digitalize and invest more in R&D activities and value added services to reduce the uncertainty related to their demand for energy.