AbstractVegetable crops production is usually based on organic fertilizers purchased off‐farm while the care of soil fertility based on the maintenance of adequate level of soil organic matter receives few attentions. Organic production in plastic tunnel‐greenhouses represents the most intensified organic production system. Therefore, we compared four alternative organic production systems mainly characterized by the combination of fertilizer application practices based on compost, cover crops and commercial organic fertilizers. The systems were: input substitution organic method (BAU), biodynamic method (BIODYN), organic‐agroecological (AGROEC) and a not fertilized but organically managed control (CNT). The objective of this study was to explore in a 3‐year long trial, the effects of different organic systems on soil organic carbon (SOC) balance and the possible effects on soil chemical–physical characteristics, enzymatic activities involved in the C, N and P cycling, total bacterial biomass (16S rRNA), and the microbial functional genes cbbL and amoA in bulk soil and in three aggregate size fractions (macroaggregates >250 μm, free microaggregates 250–53 μm and free silt plus clay (free‐SC) < 53 μm). The conducive microclimatic conditions for soil microbial activities under the greenhouse, together with the transition toward a more intensive rotation and tillage, determined a decrease of the SOC stock (from 5.9 Mg C ha−1 in BIODYN to 10.8 in CNT) that was only partially balanced by C input distributed to soil (from 4.6 Mg C ha−1 in CNT to 31.6 in AGROEC). The different organic systems did not seem to directly affect chemical and microbial properties in the bulk soils. The higher SOC decline in CNT, however, resulted in the degradation of soil structure but at same time led to the increase of the FDAse in the bulk soil. At aggregate level, instead, more evident effects of organic systems were observed on both chemical–physical and microbial properties. The impact of organic systems on enzymatic activities differed in the different aggregate size classes while the effect on bacterial biomass and functional genes was the same in all aggregate size classes. Interestingly, the cbbL gene abundance in soil aggregates, especially in macroaggregates, was correlated positively to OC inputs (r = .82; p < .001). The amoA gene and bacterial biomass, instead, put in evidence for BIODYN, a significant decrease in soil aggregates that seemed to be related, particularly in macro and free microaggregates, to the low amount of total N content. In soil aggregates, only AGROEC compared with CNT, determined the increment of the phosphatase in macroaggregates, while BIODYN and BAU highlighted even lower values of β‐d‐glucosidase, β‐d‐glucosaminidase and phosphatase in free‐SC. In conclusion in high‐input organic systems under tunnels, the misleading notion that a high supply of OC and TN through green manuring, compost amendment etc., improves per se soil health is challenged.