" Yield gap " (Yg) is a key concept of agricultural science for identifying the room for improvement of yields through better management of the agroecosystem. in rainfed agriculture Yg is the difference between actual yield (Ya) and the theoretical water limited yield (Yw) that would be achieved if solar radiation, temperature and precipitations were the only factor limiting the crop's growth and yield. Changes in Yw over regions and years are due to climate-soil interactions that are not easily modified by crop management, whereas changes in Yg are due to limiting factors that are typically within the scope of crop management such as nutrient availability, weeds, and pests. We provide an example of yield gap estimates in semi-arid a frica, using yield and other agronomic data collected in famers' fields of Senegal in 1990 and 1991 and from 2006 to 2012. i t illustrates how contrarily to what most people would expect climate is not, on average, what most limits yields in that region: yet, actual yields are on average a quarter of water limited yield, and this is due to constraints whose reduction is technically possible albeit subject to the economic and environmental relevance of doing so. Most studies dealing with the impact of climate change on agriculture in West a frica compare Yw under present and future climate as predicted by climate models. t he magnitude of those predicted long term changes in Yw by 2050 is down to –20% in the worst scenario combining a +6°C change with a -20% rainfall change. s uch changes in water limited yields are certainly concerning, but they are remarkably small compared to the potential +390% increase that would result from closing the current yield gap. When considering yield variations observed across plots and years, and not anymore regional averages over a few years, what strikes is the stability of observed yields compared to variations of Yw. We used crop model simulations with historical series of 20 years of weather data to