This paper presents an empirical study of energy demand, in which demand for a series of energy goods (Gas, Oil Products, Coal, Electricity) is expressed as a function of various factors, including temperature. Parameter values are estimated econometrically, using a dynamic panel data approach. Unlike previous studies in this field, the data sample has a global coverage, and special emphasis is given to the dynamic nature of demand, as well as to interactions between income levels and sensitivity to temperature variations. These features make the model results especially valuable in the analysis of climate change impacts. Results are interpreted in terms of derived demand for heating and cooling. Non-linearities and discontinuities emerge, making it necessary to distinguish between different countries, seasons, and energy sources. Short- and long-run temperature elasticities of demand are estimated.