Sessile oak (Quercus petraea (Matt.) Liebl.) provenance variation was assessed in a multisite test based on traits of economic and ecological relevance in France. While climatic drivers generated genetic clines at a range-wide scale, provenance variation in France was mainly shaped by past silvicultural regimes. We developed a multitrait approach to facilitate decision-making for seed sourcing. A set of provenance clusters is proposed, supporting recommendations for plantation programmes. Context Among broadleaves, sessile oak (Quercus petraea) is likely to spread in the context of current climate change and is increasingly planted in France. Seed sourcing is of the utmost importance for ensuring plantation success and adaptation. The selection of appropriate seed sources is highly challenging when the future climate conditions of plantation areas are uncertain. Aims We aimed at identifying drivers of provenance variation in Q. petraea and to build provenance clusters based on traits of adaptive and economic value, to ultimately support decision-making in seed sourcing. Methods We analysed a multisite provenance test established 30 years ago and comprising a large collection of Q. petraea provenances by performing phenotypic assessments of survival, growth, phenology, and stem-quality traits. We analysed climate-trait correlations at a range-wide scale and used multivariate statistics [multivariate mixed models, principal component analysis (PCA)] and classification methods [hierarchical clustering analysis (HCA), K-means method] to generate an overall clustering of french provenances. Results Provenance effects were highly significant regardless of the trait considered, whereas interaction effects between provenance and other experimental sources of variation were minor compared to provenance and environmental variance. There was limited variation between provenances collected in the same forest in comparison to origins of different forests. We found sharp temperature-driven genetic clines for growth and phenology-related traits at a range-wide scale. The multitrait classification approach grouped the French provenances into 11 clusters, with the members of each cluster having similar trait values. Overall, the cluster composition of provenances poorly matched the provenance regions identified on ecological grounds, but rather mirrored the silvicultural regimes implemented in the source stands in the past. Conclusions Two of the provenance clusters (comprising a total of 34 provenances) were identified as potentially useful sources of reproductive material. We recommend mixing seeds of different provenances from a given cluster to ensure the maintenance of diversity and to enhance adaptability to future climatic conditions.< Réduire