While AMMA climate scientists seek to better understand West African (WA) rainfall season, in order to be able to generate useful forecasts, agronomists objectives were to assess farmers yields forecasting methods. According to those objectives more than 2000 staple food cereals farmers fields (millet, sorghum and maize) have been monitored in Niger, Mali, Senegal and Burkina, during 2 to 5 years, with a similar methodology allowing to assess both yields variations mainly due to climate (inter-villages variability) and those due to local factors (intra village variability). Agronomical studies were also carried out in order to assess farmers varieties development features (temperature-development relationships, biomasses repartitions, photoperiod sensitivity. The studies generated a unique actualised WA farmers yields and related information database. They clearly showed that non-intensified conditions (no fertilizer, no herbicide, own seeds) are largely prevalent, actual yields are much lower than potentials, and yields variability is generally huge within a village due to effects and interactions of many factors (heterogenic manure application, micro-topography, weeds, enemies). Yet, most of the time it is not possible to statistically establish main practices effects. In this non-intensified context, climate only accounts for 30% to 60% in yields variations, and some comparisons between observed and simulated yields generated by crop-model are disappointing, when others are good. There is no real surprise in those results, and beyond the precious database, the studies provided interesting elements. They stressed the importance to distinguish between non-intensified (actual) conditions and intensified (potential) ones in prospective studies, clearly indicated it is crucial to correctly integrate traditional crops features in crop models, for both non-intensified and intensified conditions, and outlined some points that must be improved (grain-biomass relationship,