The cumulative sum (cusum) technique was examined for its use in a disease surveillance system for detecting temporal clusters of events. Optimal technique parameters were derived for scenarios not previously considered. Simulation modeling produced results that evaluated deviations from predefined rate increases. The cusum technique was less prone to false alarms and more efficient at detecting large rate increases than previously reported. As demonstrated using data obtained from a Salmonella surveillance system operated by a state animal diagnostic laboratory system, the cusum technique could provide early warning of an epidemic problem.