Electricity demand forecasting is becoming an essential tool for energy management, maintenance scheduling and investment decisions in the future liberalized energy markets and fluctuating fuel prices. To address these needs, appropriate forecasting tools for the electricity demand in Greece have been developed and tested. Electricity demand depends on economic variables and national circumstances as well as on climatic conditions. Following the analysis of the time series of electricity demand in the past decade, two statistical models have been developed, one providing daily and the other monthly demand predictions, to estimate medium term demand up to 12 months ahead, utilizing primitive (relative humidity) and derived (heating and cooling degree-days) meteorological parameters. Autoregressive structures were incorporated in both models, aiming at reducing serial correlation, which appears to bias the estimated effects of meteorological parameters on electricity demand. Both modeling approaches show a high predictive value with adjusted R-2 above 96%. Their advantages and disadvantages are discussed in this paper. The effect of the climatic conditions on the electricity demand is then further investigated via predictions under four different scenarios for the weather conditions of the coming year, which include both normal and recently observed extreme behavior. (c) 2005 Elsevier Ltd. All rights reserved.