An agricultural experimental setup has been constructed with the aim of assessing the impact of soil conservation practices on surface runoff and water quality. The site is located at Saint-Lambert-de-Lauzon (near Québec City, Canada) and is composed by twelve 624 m ² catchments for which surface and tillage runoffs, water quality (suspended matter, phosphorus, nitrate-nitrite, dissolved metals), soil physical and chemical properties, and crop yields are monitored. The experimental design allows the comparison of four agricultural treatments: two compaction treatments (with and without soil compaction) and two conservation regimes (conventional and soil conservation agricultural practices), each regime being duplicated three times. Generalized Additive Mixed Model (GAMM) highlighted significant relations between the conservation regimes and suspended matter charges, and surface runoff. In other words, conservation practices allow a significant short-term reduction of suspended matter at the field scale. On the other hand, they appear to favour an increase of surface runoff in the springtime. Since only one three-year rotation cycle has been conducted, no effect was observed on soil properties, and crop yields. Long term impacts of soil conservation practices were estimated by implementing a physically based hydrologic model SWAT over each catchment. Restored soil properties were scenarized using measurements conducted over surrounding unperturbed sites. Modelling results suggest that a restoration of soil physical properties would translate into a moderate decrease surface runoff (-5%) at the field scale. The study brings an advanced and multidimensional understanding of the field-scale processes driving soil health, quantitative hydrology, and water quality. It also quantifies potential long-term benefits of implementing soil conservation practices.