Facing health, environmental, economic and social issues, urban agriculture is currently on the rise in cities of industrialized countries. Among the different forms of urban agriculture, community gardens are characterized by their multi-functionality (source of food, education, social link…) and were identified as promising to promote healthy lifestyles (e.g. increased fruit and vegetable consumption, increased physical activity). However, only few studies are conducted in a European context, and evidence tends to be based on cross-sectional designs which do not allow causality assessment. A recent observational study conducted in the city of Marseille, France, suggested that having access to a community garden could encourage socio-economically disadvantaged women to adopt dietary practices that more closely meet dietary recommendations. Although this investigation highlighted valuable information on the role of gardening in food practices in a French context, causality still could not be inferred due to a transversal design. Moreover, given the widening gap in socio-economic inequalities in health in Europe, including in France, more research is needed on the possible differential impacts of such experiment on individuals, depending on their socioeconomic positions and incomes. The objective of this study is to design a protocol for a quasi-experimental evaluation allowing to assess the impact of access to a community garden on nutritional quality and sustainability of food practices, and if this impact varies between low- and middle-income populations. Methods/Design: This study will be based on a natural experiment (access to a community garden) evaluated within a quasi-experimental design. Gardeners who recently had access to a community garden will be recruited in two socio-economically contrasted neighborhoods of Montpellier, France, to constitute a low-income and a middle-income natural experiment groups. Volunteers living in the same neighborhoods but with no access to a community garden will be recruited as comparison groups (control). Data collection at recruitment (baseline) and at follow up at 1 year (endline) will include household monthly food supply (food purchase receipts, garden productions, gifts, food aid), food waste behavior, and socio-demographic characteristics. Qualitative approaches will be used to explore intermediate variables on a causal path between access to the garden and food practices changes, such as gardener’s motives for sustainable food systems and sustainable food choices, length of time with access to a garden, personal history of gardening, or self-efficacy. The primary outcome will be the sustainability of food practices, assessed through the nutritional quality and environmental impact of household’s food purchases, the later being estimated through food waste, diet-related greenhouse gas emissions and the animal-to-plant ratio. This protocol is part of the SURFOOD-Foodscapes project (Sustainable Urban Food Systems – The effects of urban foodscape on food styles in Montpellier Metropole). Discussion: Change of outcomes after 1 year will be compared between the natural experiment and control groups, to evaluate the effect of access to a community garden on the sustainability of food practices. Differences induced will be compared between the low- and middle-income groups in order to assess the impact on social inequalities in health. This study will provide valuable information for urban strategies to favor healthy and sustainable food practices, based on a robust quasi-experimental design allowing causality evaluation.