A new method is proposed for forecasting electricity load-duration cur
ves. The approach first forecasts the load curve and then uses the res
ulting predictive densities to forecast the load-duration curve. A vir
tue of this procedure is that both load curves and load-duration curve
s can be predicted using the same model, and confidence intervals can
be generated for both predictions. The procedure is applied to the pro
blem of predicting New Zealand electricity consumption. A structural t
ime-series model is used to forecast the load curve based on half-hour
ly data. The model is tailored to handle effects such as daylight savi
ngs, holidays and weekends, as well as trend, annual, weekly and daily
cycles. Time-series methods, including Kalman filtering, smoothing an
d prediction, are used to fit the model and to achieve the desired for
ecasts of the load-duration curve.