Summer leguminous cover crops can improve soil health and reduce the economic and environmental costs associated with N fertilizers. However, adoption is often constrained by poor weed suppression compared to nonlegume cover crops. In field experiments conducted in organic vegetable cropping systems in north-central New York, two primary hypotheses were tested: (1) mixtures of legume cover crops (cowpea and soybean) with grasses (sorghum–sudangrass and Japanese millet) reduce weed seed production and increase cover crop productivity relative to legume monocultures and (2) higher soil fertility shifts the competitive outcome in favor of weeds and nonlegume cover crops. Cover crops were grown either alone or in grass–legume combinations with or without composted chicken manure. Under hot, dry conditions in 2005, cowpea and soybean cover crops were severely suppressed by weeds in monoculture and by sorghum–sudangrass in mixtures, resulting in low legume biomass, poor nodulation, and high levels of Powell amaranth seed production (> 25,000 seeds m−2). Under more typical temperature and rainfall conditions in 2006, cowpea mixtures with Japanese millet stimulated cowpea biomass production and nodulation compared to monoculture, but soybeans were suppressed in mixtures with both grasses. Composted chicken manure shifted competition in favor of weeds at the expense of cowpea (2005), stimulated weed and grass biomass production (2006), and suppressed nodulation of soybean (2006). In a complementary on-farm trial, cowpea mixtures with sorghum–sudangrass suppressed weed biomass by 99%; however, both common purslane and hairy galinsoga produced sufficient seeds (600 seeds m−2) to replenish the existing weed seedbank. Results suggest that (1) mixtures of cowpeas with grasses can improve nodulation, lower seed costs, and reduce the risk of weed seed production; (2) soybean is not compatible with grasses in mixture; and (3) future costs of weed seed production must be considered when determining optimal cover crop choices.