In France, reports of spatio-temporal clusters of non-infectious diseases within the general population have been rising sharply since 2000. The severity of the reported diseases (often cancers) and the presence of potential risk factors in the general environment are causing concern among the affected populations and generating an urgent demand for intervention. An analysis of the history of investigations into spatio-temporal clusters reveals limited success in identifying an etiological factor for these events, despite the frequent commitment of significant resources. The main reasons for this are the frequency of random clusters of diseases, as well as the methodological biases and difficulties encountered when investigating a small number of cases of often diverse and multifactorial conditions. To provide an appropriate public health response that allows both for conducting the investigation with the necessary scientific rigor and for conserving public resources, it is proposed to adopt a specific investigation protocol. This protocol is based on a prioritized collection of health and environmental data and on a stepwise analysis of the plausibility of an aggregate resulting from a common exposure among populations. Given the social climate surrounding these events, particular attention is paid to relations with the public