Every year wildfires are responsible of the loss of thousands of hactares of European forest, millions of dollars in damage and, in the worst case, for loss of human lives. Accurate observation of wildland fire evolution is a crucial issue to estimate the burned area, to use the observed data in propagation prediction and to calibrate input parameters of propagation models. In this direction the European Forest Fire Information System (EFFIS) supports the services in charge of the protection of European forests. Satellite images provide useful information, but, in many cases, the clouds or the fire smoke itself do not allow to have a good estimation of the fire front position. So, a methodology based on thermal anomalies information has been applied to estimate fire perimeters even on adverse conditions. This methodology has been recently added to EFFIS to determine the perimeter evolution of a wildfire using thermal anomalies information coming from NASA's satellites. In this paper we describe the algorithm to perform this task and how it was implemented and integrated into the EFFIS services.