Health insurance databases are the only resources available in France that can be used for the online detection of clusters of acute gastroenteritis. It is possible to subsequently design an automated procedure to assess the likelihood of waterborne causes for identified clusters by comparing the geographical extent of the cluster with water distribution networks. Thirty to 60% of gastroenteritis cases that have been the subject of a medical consultation in France would be available within a timeframe compatible with the alert objective (3 business days). However, consultations conducted outside the insured person’s place of residence would not be usable. The insured person’s address is available, but locating cases by municipality would simplify the analysis process and, in principle, would have little impact on the system’s performance. These initial conclusions must be tested in the field. To this end, a list of departments eligible for the pilot program has been drawn up. The size category of the target distribution units was set between 500 residents served—the statistical power limit of the alert system—and 10,000, the threshold above which microbiological contamination is too rare to draw robust statistical conclusions. The occurrence of "critical events"—defined as the detection of more than 50 colonies of fecal bacteria (Escherichia coli and enterococci) in a sample collected for regulatory testing—was selected as the indicator of epidemic risk. The criteria for ranking the departments were (i) the number of distribution units in the population class served [500–10,000] affected by at least one critical event during 2003 and (ii) the size of the population concerned. The overseas departments (Réunion, Guadeloupe) and rural or karstic departments (Cantal, Charente-Maritime, Doubs, Lot, etc.) rank highest based on 2003 data. High-mountain departments do not appear in the ranking because water distribution there is highly fragmented, making an epidemiological approach difficult. Regardless of the size of the distribution systems of interest and the sampling effort, a "departmental" effect appears to emerge, predisposing Cantal, Lot, Réunion, and Doubs to a higher risk of microbiological contamination likely to generate "observable" clusters than elsewhere. Eligible DDASS offices and the Health Insurance Fund will be consulted regarding the selection of pilot departments, data collection, and the formation of the partnership. (R.A.)