A method is described for detecting load changes and characterizing subsequent responses in single-input-single-output (SISO) feedback control loops. The method is computationally simple and suitable for on-line implementation in a controller. Load changes are detected using a statistical change detection procedure that uses zero-crossings information to account for autocorrelation. Responses are characterized by a normalized index that can be related to a damping ratio in a second-order model. Results are presented from applying the method to data from a simulated control loop and two temperature control loops in a large office building. (c) 2004 Elsevier Ltd. All rights reserved.