Improved public health surveillance can lead to earlier implementation of prevention and control measures. Better surveillance data lead to a more rational establishment of priorities. More timely and accurate data facilitate earlier epidemic detection and control. With better surveillance data, the impact of intervention activities and other public health programs can be evaluated more accurately. In this paper we describe how to improve the science of surveillance in terms of data collection, analysis, and dissemination and its application to public health practice. We then discuss the potential benefits and costs of such efforts and suggest methods for evaluating alternative approaches. The argument for science in surveillance, on the other hand, may be subject to excess. Surveillance is not an end unto itself, but rather a tool. This tool should be refined and modified to adapt to the goals of a particular public health program. It is the development of methods to apply creative ideas to surveillance, and the rigorous assessment of the process, that will benefit from the application of scientific principles.