The document summarizes a study comparing time series and artificial neural network (ANN) methods for short-term load forecasting of Covenant University, Nigeria. Load data from October 15-16, 2012 was used to develop forecasting models using moving average, exponential smoothing (time series methods) and ANN. The ANN model with inputs of previous load, time of day, day of week and weekday/weekend proved most accurate with a mean absolute deviation of 0.225, mean squared error of 0.095 and mean absolute percent error of 8.25, making it the best forecasting method according to the error measurements.