In traditional agricultural areas, where traditional crops (TCs) are cultivated, small farms are still highly represented. Located prevalently in marginal and sensitive areas, agricultural areas have undergone deep transformation. Smallholders have maintained the traditional asset of cultivation (extensive and low input requirement management) only to some extent. In some cases they have adapted traditional orchards into more intensive planting systems. Frequently, they have abandoned agriculture. The land use and management influence soil functions, i.e., the capability of a specific soil to provide key functions in terrestrial ecosystems. In order to assess whether small farms are environmentally sustainable, we used a set of soil quality indicators in three traditional tree crops in the Latium region (central Italy), like hazelnut, grapevine, and Citrus. The soil parameters, chemical, biological, and biochemical, were quantified under three different management models: extensive cultivation, intensive cultivation, and abandonment. The selected set of indicators proved to be able to discriminate adequately between the management models and to be suitable for the soil health assessment. Results proved that hazelnut orchards stored more organic C, independently from farming management, while vineyard showed the lower total organic carbon (TOC). The microbial carbon vs. organic carbon ratio (Cmic-to Corg ratio) was higher for vineyards and Citrus groves, denoting a more active degradation of soil organic matter. Soil enzymes (ESs) involved in C cycle were variable along the different treatments and mainly influenced by the C inputs to soil and soil cover, whereas those involved in N, P, and S cycles were higher in abandoned and extensive TCs. Overall, extensive cultivation performed better in terms of soil quality than intensive or abandonment. This study suggests that a transition to an agriculture based on agroecological principles and toward extensification would provide significant soil-based environmental benefits in marginal sensitive areas.