Aims: We investigated whether drought-induced impairment of grassland species can be explained directly by plant water deficit or by water-driven limitation of nitrogen (N) and/or carbohydrate sources. Methods: In a field experiment, a severe drought treatment was applied on monocultures of Lolium perenne L. (cv. Alligator) (Lp) and Trifolium repens L. (cv. Hebe) (Tr) by using rainout shelters excluding all precipitation, and effects were compared to a rainfed control. Three species-fertiliser treatments were set up, crossed with the drought treatment. The two species were fertilised equally with N (200 kgN ha−1 year−1), and an additional high N fertilisation treatment was established for L. perenne (LphighN, 500 kg N ha−1 year−1). Results: Severe soil water deficit led to significantly lower leaf water potentials in all species-fertiliser treatments (P < 0.001) down to approximately −1.2 MPa and, on average, to a 79% reduction in living plant biomass above 7 cm harvest height (P < 0.001), indicating strong plant water deficits. Under the drought treatment, living plant biomass above 7 cm did not differ among species-fertiliser treatments. Plant-available soil N was 84% lower (P ≤ 0.01) and plant N concentrations were 24% less (P < 0.001) under the drought than under the rainfed control treatment, with Lp always being more N limited than LphighN and Tr. Nitrate concentrations in water-limited plants were generally very low (< 0.85 mg g−1 dry matter), whereas non-structural carbohydrates were distinctly greater under the drought treatment in Lp (+62%), LphighN (+46%), and Tr (+18%). Conclusions: Restricted biomass production of these forage species under severe drought can primarily be explained by plant water deficits and secondarily by drought-induced limitation of N supply. However, growth seems not to be limited by carbohydrate source activity, as carbohydrates accumulated with water deficiency.