To predict and respond to famine and other forms of food insecurity, different early warning systems are using remote analyses of crop condition and agricultural production, using satellite-based information. To improve these predictions, a reliable estimation of the cultivated area at national scale must be carried out. In this study, we develop a methodology for extracting cultivated domain based on their temporal behaviour as captured in time-series of moderate resolution remote sensing MODIS images. We also used higher resolution SPOT and LANDSAT images for identifying cultivated areas used in training. We tested this methodology in Senegal and Mali at national scale. Both studied areas were stratified in homogeneous areas from an ecological and a remote sensing point of view, to reduce the land surface reflectance variability in the dataset in order to improve the classification efficiency. A spatiotemporal (K-means) classification was finally made on the MODIS NDVI time series, inside each of the agro-ecological regions For Senegal, we obtained an updated map of crop area with a better resolution than the USAID map (which is 1 km resolution) and with a nomenclature more specific of the Senegal region than suggested in the POSTEL map. For Mali, the results showed that MODIS data set can provide a completely satisfactory representation of the cultivated domain in one FEWS zone, in combination with external data. Results at national scale are being processed and will be presented at the conference.