The predictability of northern Africa dust events is assessed using daily numerical forecast simulations for the next 3 days. The dust concentration fields, modeled with the CHIMERE-DUST model, were first evaluated by comparison with both Aerosol Robotic Network (AERONET) surface data and Ozone Monitoring Instrument (OMI) and Spinning Enhanced Visible and Infrared Imager (SEVIRI) satellite measurements. The accuracy and spread between measurements and simulations are discussed for the first short observation period of the African Monsoon Multidisciplinary Analysis (AMMA) experiment in western Africa, between January and March 2006. The predictability of dust events was then estimated by comparing model results for different leads in a forecast mode. The model performance was evaluated with respect to its capability to forecast the surface wind speed, which is the key process for dust emission, and the transport of mineral dust near source regions and toward remote areas. It is shown that forecast emissions can vary up to 80% (close to the sources) but that the variability on forecasted dust concentrations and optical thicknesses do not exceed 40% and 20%.