Forecasting precipitation in the western Mediterranean is difficult because of the interactions among dynamical forcing, orographic lifting and moisture advection from the warm Mediterranean Sea. Torrential rainfall events are not uncommon, especially during the autumn. This type of event motivated an effort to improve precipitation forecasting by incorporating additional information on the initial state of the humidity field from Global Positioning System measurements of refractive delay. In this study we process data from a network of sites in Western Europe and assimilate the data over a two week period into the HIRLAM numerical weather prediction model. The overall impact for the two week period is neutral, however, for a severe rain event taking place during that period, the forecasts show improved skill when including GPS data. The work implies that the GPS data have good potential for influencing numerical models in rapidly developing, high moisture flux situations.