This research focuses on developing methodologies for building load analysis and forecasting. A set of one-year, 15-min energy consumption data, collected at a substation feeder supplying the Centennial Campus load at the NC State University, is used in the study. The correlations of building load consumption with respect to the outdoor temperature, type of the day, time of the day, and humidity are studied. Then, four two-hour-ahead forecast methods are developed. The performance of each method in terms of accuracy, robustness, and efficiency is compared. The contributions of the research are 1) the building energy signatures allow building owners to identify abnormal operation scenarios and quantify the status of building operation states; 2) the load profile analysis helps building owners to decide how to optimize the building operation by shifting consumptions; and 3) the load forecasting methods predicting future operation conditions help building owner to take precaution on abnormal operation conditions to avoid economic losses and improve operational reliability.