Crop establishment is a critical step for optimal agronomie and economic crop performances. In sunflower, average emergence rate stands between 70-85% in France, with a tendency to decrease because of increasing bird predation. Consequently, farmers have to sow again the crop in about 8% of the situations. But obtaining a successful plant stand depends not only on the crop emergence rate but also on the initial sowing density. Currently, the practical recommendation for sowing density is between 65-70k seeds per ha (target of 50-60k plants/ha) which should be increased (70-75k seeds/ha) in case of minimum tillage, because oflower crop emergence risk. Ideally sowing density should also be adapted to expected soi! water resources, pathogen pressure in the growing area, and cultivar. Our study aims to refine the current sowing recommendations from Terres Innovia by providing site-specific decisions based on climate, soi!, and cultivar. We used field experiments (multi-environment trials) to assess yield and oil concentration response to plant density, soi! type, and cultivar in French growing conditions. We then extended this dataset with numerical experiments (simulations from SUNFLO crop mode!) to evaluate how climatic uncertainty and crop management were impacting or not current sowing recommendations. Datasets were analyzed with linear and quantile regressions both globally and for subsets of the studied population of crops. We found that crop yield variance was mainly explained by farm location (soi! and soi! x climate interactions), with a weak average impact of plant density (<= 5%); the hierarchy of factors was similar with field or simulated dataset. When refining practical recommendations to account for available soi! water content (A WC), adjustment of plant density had a greater impact on crop performance. In deep (A WC > 200 mm) and intermediate (1 Oümm