In August 2003, a heat wave with exceptional health consequences struck France. Its intensity, duration, and scope reflect the observed impact. The health crisis led, on the one hand, to massive use of the emergency care system and, on the other hand, to a sharp increase in mortality. The first noticeable consequence was the saturation of the healthcare system, while health surveillance services and networks—which lacked warning indicators—failed to respond as expected. This health crisis necessitated a review of the health surveillance systems designed for early warning that were available in France at the time. At that time, public health surveillance was organized almost exclusively around disease-specific systems (mandatory reporting, sentinel networks), which were not adaptable to unforeseen events. This crisis demonstrated that such phenomena—in terms of their origin, geographic scope, and consequences—do exist and will therefore recur. This does not mean that the “disease-specific” systems currently in place lack relevance. On the contrary, they are highly effective and have enabled the early identification of numerous diseases—most often infectious—and have led to coordinated management programs to limit their health impact. Following the 2003 heatwave, a public health action plan was implemented in anticipation of a new heatwave. However, the value of a public health surveillance system based on non-specific health data—such as the flow of patients seeking care through emergency medical systems—was also emphasized. A system was developed by the InVS based on a hospital network, community-based care, and mortality records. Its objectives are to identify health situations requiring an appropriate public health response, as well as to measure the impact of epidemics or anticipated events. This article describes the experimental organization of this surveillance system in France.