Below ground carbon (BGC) inputs to soil, i.e. root biomass and rhizodeposition carbon (C), are among the most important variables driving soil C dynamics in agroecosystems. Hence, increasing BGC inputs to deep soil is a proposed strategy to sequester C in the long term. As BGC inputs are inherently difficult to measure in the field, they are usually estimated from yield in order to supply soil C models with input data. While fertilization intensity considerably affects above ground biomass, its influence on BGC inputs is largely unclear, especially with respect to the subsoil. Therefore, we determined net root biomass and rhizodeposition C of field-grown maize and wheat at harvest in different farming systems (bio-organic, conventional) and fertilization treatments (zero, manure, mineral) along an intensity gradient in two Swiss long-term field trials. Plants in microplots were repeatedly pulse-labelled with 13C-CO2 throughout the growing seasons and shoots, roots, and soil to 0.75 m depth were sampled at harvest. Despite a strong increase of above ground biomass with increasing fertilization intensity, BGC inputs were similar among treatments on both sites irrespective of soil depth. However, the proportions of rhizodeposition C of BGC inputs averaged 54 to 63% and were, therefore, much larger than the widely adopted 40% for field-grown cereals. They increased with soil depth and were highest under sole organic fertilization. The shift in whole-plant C allocation towards above ground biomass with increasing fertilization intensity entailed 10% higher C allocation below ground in organic than conventional farming for both maize and wheat. Our findings imply that yield-independent values provide closer estimates for BGC inputs to soil of cereals in different farming systems than yield-based functions. We further conclude that fertilization has only little potential to alter absolute amounts of BGC inputs to deep soil in order to sequester C in the long term.