Introduction Nitrogen (N) is an important factor for yield and baking quality of wheat. However, due to environmental concerns, N inputs shall be reduced in Switzerland while at the same time reduced water availability due to climate change hinders N uptake. Due to the relationship between N input and baking quality, N reduction constitutes a challenge and necessitates varieties securing high and stable baking quality. To breed such varieties, an in-depth understanding of baking quality and advanced prediction methods in early breeding generations are crucial. Objective In this project, we will use existing datasets from two breeding programs, variety registration and recommendation as well as downstream wheat production and processing. We aim to establish and predict efficiency indices which indirectly relate quality traits to N input, thus turning available N more efficiently into baking quality. To improve the selection of such N efficient lines, we will develop prediction methods of these derived indices and final baking quality from a reduced set of traits, genomic data, protein composition properties, using regression methods, genomic selection, and machine learning models. Expected outcomes and results These advanced prediction methods will lead to more efficient selection, accelerated genetic gain and consequently the release of varieties with high baking quality that better use available N more efficiently. This will contribute to a more sustainable wheat production and secure self-sufficiency of Swiss wheat cropping.