To investigate the impact of integrating ley grassland in crop rotations on soil C changes at depths of 0-30 and 0-60 cm, field data regarding C input, soil C and crop production were collected during 2005-2016 from a long-term experimental site in Lusignan, southwestern France. Root biomass C was sampled annually, and the stable C isotope signature (delta C-13) was determined to quantify the amount of C input from the root biomass. The results showed that integrating ley grassland in crop rotations increased the C input in the 0-30 and 0-60 cm soil layers. Carbon stocks also increased in the 0-30 cm soil layer but not always in the 0-60 cm layer. However, during the first full rotation (2005-2013), treatments involving 3-year and 6-year grasslands showed 37.5% and 62.5% losses of their crop production, respectively. In crop rotations, integrating ley grassland together with high-nitrogen (N) or low-N fertilisation did not impact the C input, changes in soil C in either the 0-30 or 0-60 cm soil layer, or even crop production, while grass production decreased by 22.2-66.6% under low-N fertilisation. Compared with cropland, permanent grassland did not show a greater C input, whereas the latter showed a C stock increase of 0.6-1.4 Mg C ha(-1) yr(-1). The analysis of 2005-2013 crop rotation data suggested that both above- and belowground C inputs should be considered as a single total C input. An estimated baseline C input rate of 1.9 Mg C ha(-1) yr(-1) was suggested to maintain soil C stocks for some of the crop rotations at the site. Considering the balance of food production and soil C stocks, crop rotation in conjunction with 3 years of high-N-fertilised grassland is recommended.