Crop rotations effects are key features of environmentally friendly crop production practices but they are poorly documented. The aim of this article is to present an estimation approach of these effects based on farm accountancy data. Estimating crop rotation effects on yield and variable input use levels from farm accountancy data is a challenging issue since farmers crop sequence choices are not observed. We propose to devise this estimation problem as a Bi-level Programming (BLP) problem designed to estimate the crop rotation effects while simultaneously reconstructing farmers unobserved crop sequences from farmers observed (current and previous) crop acreages. Our estimation approach is based on a well-defined statistical background. It relies on simple crop sequence yield and input use models as well as on an assumption stating that farmers are economically rational when deciding their crop sequence acreages. Our estimation approach also makes use of expert knowledge information on crop rotation effects for guiding the construction of the crop rotation effect estimators. An illustrative application based on French farm accountancy data demonstrate that the proposed estimation approach yields meaningful crop rotation estimates when suitably implemented. Acknowledgement : This research has benefitted from financial support from the European Commission.