This study aimed to study the relationship between milk quality for cheese manufacturing and farm's environmental impacts in grass-based mountain systems. Beforehand, the multi-criteria evaluation methodology proposed by Botreau et al. and Rey-Cadilhac et al. was adapted to a routine use. Milk quality scores were constructed by weighing traits possibly predicted by spectroscopic analyses currently available in milk labs into 4 dimensions: health, nutritional, technological and sensory. Environmental impact scores were constructed by combining 6 indicators of the CAP'2ER (R) software in 5 dimensions: greenhouse gas emissions, eutrophication, air acidification, consumption of space and non-renewable energies and ecosystem biodiversity. The study sample included 15 dairy farms located in the French Massif Central mountains. An analysis of variance was carried out to study the effect of system's types on milk quality and environmental impact scores. A principal component analysis was used to study the relationship between milk quality and environmental impact dimensions scores. No correlation has been established between overall milk quality and overall environmental impact scores. High scores for the nutritional and biodiversity dimensions and low scores for the resource consumption dimension, were associated with farming practices such as a high proportion of grasslands in the usable agricultural area, low stocking rate and low productivity per cow. This demonstrates the importance of defining specific priority objectives, on a farm-by-farm basis, in order to drive changes in agricultural practices. The multi-criteria evaluation model tested here appeared sensitive, but it needs to be tested on a larger scale and in different contexts.