Enzymes are known to be sensitive indicators of soil quality. Nevertheless, the natural variability of their activities needs to be considered for a relevant interpretation of activity levels, especially in contaminated soils. We first monitored the variability of enzymatic activities (acid and alkaline phosphatases, β-glucosidase, N-acetyl-β-glucosaminidase, urease and dehydrogenase) in two agricultural sites in north-western France during a seasonal April–October cycle. For both sites, two types of long-term different agricultural managements, grassland and intensive cropping were considered. Our results revealed great variability of enzymatic activities in space and over time, yet more pronounced for grassland than for cropped soils. Then, we assessed the impact of copper on enzymes activity on Terrestrial Model Ecosystems (TMEs) filled with undisturbed soil, and incubated for 70 days in open-field conditions. Copper was added at two concentration levels corresponding to a regulated annual agronomic input (2 mg kg−1) or to a high soil contamination (200 mg kg−1). In comparison to effects of natural spatiotemporal variability of soil conditions, copper addition did not show significant impacts on enzymatic activities. Finally, our results confirmed that for assessing effective impacts of contaminants in soils under real field conditions, natural spatiotemporal soil variability must be considered.