Spatial variations in health indicators and environmental exposure factors are studied in epidemiology for descriptive purposes and to suggest etiological hypotheses. The objective of this paper is to present and discuss the main tools and methods that utilize geographic information systems (GIS) and spatial statistics in ecological-geographic studies. This work focuses on ecological studies in which variables (health indicators and risk factors) are measured at the scale of a geographic unit (municipality, statistical information clusters (Iris), etc.) rather than at the individual level. Statistical methods suited to analyzing the relationships between health indicators and indicators of exposure to environmental risk factors are described and discussed. The modeling and statistical analysis of these data pose a number of methodological challenges: high variability, spatial dependence, the existence of different spatial scales, etc. The most commonly used statistical tools to address these challenges are presented. The possibilities offered by the implementation of GIS and the use of geographic data are presented using concrete examples of work conducted at the Department of Environmental Health (DSE) of the French Institute for Public Health Surveillance (InVS) as well as some examples from the literature, with an emphasis on the precautions that must accompany their use. (R.A.)