In the past century, most academic literature and industrial trials in the forecasting area have focused primarily on point load forecasting. Recently, however, an increased share of intermittent energy sources in the electricity grid and the introduction of prosumers (i.e. consumers who produce electricity) have significantly increased the uncertainty in both electricity generation and demand. In today’s electricity market, new forecasting tools are therefore required to manage this inherent uncertainty. To this end, this paper develops a probabilistic forecasting framework based on Bayesian Neural Networks (BNNs) with optimised initialisation for one-day ahead forecasting of operational demand across the National Electricity Market (NEM) of Australia. Probabilistic forecasting aims to maximise the reliability and sharpness of the predictive distributions, subject to calibration, based on the available dataset. Those predictive distributions can provide insights in an uncertain environment that help energy analysts to better assess potential forecast variability.