In the early 2000s, occupational physicians at Aero, a major aerospace group, teamed up with researchers to build a new type of statistics. Embedded in the EVREST (EVolutions et RElations en Santé au Travail) scheme, these statistics rely on a questionnaire based survey administered during the medical check-up. Each year, the occupational physicians report the survey results to the social partners in several workgroups. Through the use of statistics issued from EVREST, the physicians wanted to display a link between working conditions and health issues, which was often ignored by the institution and its representation of occupational health and safety. The purpose of this article is to analyze the use of quantification, its genesis and its effects. Focussing on the "conflicting uses" around the numbers, this article demonstrates that their production is part of social relations that contribute to the redefinition of the goals initially assigned to instrument by their creators. Finally, the analysis aims to contribute to the reflection on the social conditions for the development and acceptance of alternative indicators