Aims: Belowground legume nitrogen (N) composed of roots and rhizodeposition is an important N input to soils, but published data of belowground N vary broadly, probably due to extrapolation from short-term experiments and dissimilar growing conditions. We quantified belowground N inputs of red clover (Trifolium pratense L.) during two consecutive years in a clover-grass sward along a soil nutrient availability gradient. Methods: We established a red clover-perennial ryegrass (Lolium perenne L.) model sward in microplots located in field plots of the DOK experiment, which has a 33-year history of organic and conventional cropping, resulting in a soil nutrient availability gradient. Four treatments were examined: the zero fertilisation control, bio-organic with half and full dose manure application, and the conventional system with mineral fertilisation at full dose. We studied the development of clover aboveground and belowground N using multiple pulse 15N urea leaf labelling. Results: Belowground clover N increased over time and with rising nutrient availability and was proportional to aboveground clover N at all times. Belowground clover N amounted to 40% of aboveground clover N during two consecutive years, irrespective of the nutrient availability status. Belowground clover N development was initially dominated by fast root growth, followed by enhanced root turnover during the second year. Potassium availability limited clover growth and total N accumulation in treatments with low nutrient availability. Conclusions: Belowground red clover N inputs could be estimated from aboveground N by a constant factor of 0.4, regardless of the nutrient availability and cultivation time. Root turnover led to a distinct absolute increase of N rhizodeposition over time. Hence, N rhizodeposition, with an 80% share of belowground N, was the predominant N pool at the end of the second year. © 2018, Springer International Publishing AG, part of Springer Nature.