Meat sheep production in most countries relies on grazing and profitability (gross margin per ewe and per year) is low. Profitability is in part based on the level of numeric productivity (number of lambs produced per ewe and per year). Numeric productivity depends on fecundity and survival rates of ewes and lambs. These survival rates are extremely variable from one farm to another. These variations are detrimental to sustainability of the farms. We will study the mortality rates of ewes and lambs and other health related traits, taken as indicator in meat sheep farms in different locations-plain or semi-mountain areas (34 farms) in relation to gross margin taken as an indicator of economical sustainability as well as other markers of sustainability. We constructed a method for evaluating sustainability based on the above mentioned parameters in the field of production, animal health, and economics. It was based on two step procedure: first, selection of farm parameters using clustering methods, second ranking farms on sustainability using principal component analysis with an orientation obtained by the incorporation in the analysis of two virtual farms (high and low sustainability) constructed on a choice of sustainability for each parameter. This choice of sustainability is flexible and dependant on the views of stakeholders and may modify the ranking of farms. The scale of sustainability we proposed was tested in relation to the management of the farm (organic or conventional) and location (plain or semi-mountain areas): the semi-mountain farms were more sustainable than plain farms and organic farms were often, but not always, more sustainable than conventional farms. The method can be easily applied to any animal production.