Farmers are increasingly looking to cover crops to provide ecosystem services to support yields, while reducing off-farm inputs. Research evidence in both unmanaged and agricultural ecosystems suggests that increased plant diversity can improve ecosystem service outcomes. The use of diverse cover crops as a practice to increase ecosystem service delivery has potential, though knowledge gaps remain regarding the types and levels of diversity that are most effective, and how environment may impact the desired outcomes. Using six species of legumes (hairy vetch, field pea, crimson clover) and of grasses (cereal rye, common wheat, ryegrass) along with multiple cultivars of each, we conducted an annual field experiment repeated twice during the winter fallow season. We tested the effect of a gradient of diversity in intraspecific mixtures, interspecific mixtures, and functional group (grass and legume) mixtures. We measured aboveground biomass production, weed biomass, soil-derived and fixed nitrogen in the shoots, and the C/N ratio of the aboveground biomass to evaluate corresponding ecosystem services relating to C and N cycling and weed suppression. Species mixtures, especially those composed of both grasses and legumes tended to have the most significant ecosystem service benefits suggesting that functional diversity is more significant than species richness. Where soil nutrients were less available, the diversity effect tended to be stronger, though this was significant in only two instances. These results lend some support to the stress-gradient hypothesis in that in more stressful conditions the diversity benefit was greater. Diverse mixtures may not consistently produce substantial service improvements, but there is little risk, and they may be most useful in marginal or depleted field conditions.