The scopus of this paper is to compare the performance of different structures of Artificial Neural Networks (ANNs) regarding the input variables used for short-term forecasting of the next day load in intercontinental Greek power system. The input variables can be: (a) historical loads, (b) weather related temperatures, (c) hour and day indicators, in two ways: (i) selfsame, (ii) compressed using the Principal Components Analysis (PCA). The training algorithm is the scaled conjugate gradient algorithm, for which a calibration process is conducted regarding the crucial parameters values, such as the number of neurons, the kind of activation functions, etc. The performance of each structure is evaluated by the Mean Absolute Percentage Error (MAPE) between the experimental and estimated values of the hourly load demand of the next day for the evaluation set in order to specify the optimal ANN. Finally the load demand for the next day of the test set (with the historical data of the current year) is estimated using the best ANN structure, so that the verification of behaviour of ANN load prediction techniques was demonstrated.