We show that ecosystem-specific aboveground net primary productivity (SANPP, g g(-1) day(-1), productivity on a per gram basis) can be predicted from species-level measures of potential relative growth rate (RGRmax), but only if RGRmax is weighted according to the species' relative abundance. This is in agreement with Grime's mass-ratio hypothesis. Productivity was measured in 12 sites in a French Mediterranean post-agricultural succession, while RGRmax was measured on 26 of the most abundant species from this successional sere, grown hydroponically. RGRmax was only weakly correlated (r2 = 0.12, P < 0.05) with field age when species abundance was not considered, but the two variables were strongly correlated (r2 = 0.81, P < 0.001) when the relative abundance of species in each field was taken into account. SANPP also decreased significantly with field age. This resulted in a tight relationship (r2 = 0.77, P < 0.001) between productivity and RGRmax weighted according to species relative biomass contribution. Our study shows that scaling-up from the potential properties of individual species is possible, and that information on potential and realized species traits can be integrated to predict ecosystem functioning.