In winter crops, leaf area is a major determinant of the final yield, and is substantially affected by losses occurring during vegetative growth. Here, we propose and test a submodel simulating the development of leaf area and pod area, along with leaf senescence, for winter oilseed rape (Brassica napus L.), which was included in a CERES-type model for rape adapted from CERES-N Maize. This crop model, called CERES-Rape, has components for crop phenology, net photosynthesis, N uptake, and assimilate partitioning. As a new feature compared to previously published work, che leaf area submodel includes senescence from shading due to competition for light in the canopy, and from leaf N deficiencies. The model has been developed and parameterised on a 1-yr-long experiment with three fertilizer N treatments in northeastern France, during which measurements of senescing parts allowed calibration of the equations for leaf area index (LAI) senescence and total generated LAI. The leaf area submodel, once coupled to the CERES-Rape model, was tested against two additional experiments from Denmark and northern France. This process-oriented submodel proved accurate for the simulation of actual LAI whether in the calibration or in the validation phase, with an overall Root Mean Square Error (RMSE) of 0.496 m(2) m(-2), falling close to the mean experimental standard deviation. Extrapolation did not require any further adjustment, although a different cultivar was involved. (C) 1998 Elsevier Science B.V. All rights reserved.