Mixed farming systems are of interest in the search for sustainability because of their species diversity and the potential for synergy from integrating crops with livestock. However, their ability to maximize food production has been little addressed in the literature and deserves to be explored further. The issue of nutrient recycling raises questions about resource allocation between food crops, feed and animal products. This study, based on a whole farm system experiment conducted for approximately fifteen years in northeastern France, assesses the biotechnical processes and food production performance of two integrated mixed system configurations. These configurations differed both in their types of production (diversity in both livestock and crops) and in their overall strategies (striving for self-sufficiency vs. maximizing food-crop output). Taking a metabolic approach, the study evaluates biotechnical processes (by ecological network analysis and nutrient balances) and food production efficiency. Our results show that the configuration geared to maximizing food production is not the more productive but is the more efficient. In both cases, efficiency at the farm system scale is better than the efficiencies of each production. This confirms the importance of combining systemic and analytical approaches to better understand and act on the development of agroecological farming systems. We also show the importance, for a self-sufficient system, of having stocks in reserve to cope with unfavorable years. Finally, our study confirms the value of integrated mixed farming systems in terms of agroecology but highlights the need for (i) a closer consideration of their food production aspect and (ii) an analysis of the temporal dynamics of agrosystems and the trade-offs between food production and nutrient cycling.