Precision farming allows automatic and massive data collection aiming at improving animal productivity and welfare among many goals. Animal models are needed to interpret these data and individual variability in order to better adjust management strategies (i.e. feeding strategies). The GARUNS model of dairy cow lifetime performance initially proposed by Martin and Sauvant has evolved through the last ten years, including a reproductive module, able to take into account individual reproductive management, and more recently a ‘feed plan’ module, taking into account the composition of the ration distributed to the cow during its life. The first objective of this study was to quantify the effect of this ‘feed plan’ module on the fitting accuracy of production variables. The second objective was to evaluate the fitting quality of the second lactation when the fitting was done on first lactation data only. The new model, with the ‘feed plan’ module, was fitted with a step-by-step procedure on the first lactation data of 16 Holstein cows. These data were collected during an experimental trial on extended lactation conducted from 2012 to 2015 at the Danish Cattle Centre in Aarhus University (Denmark). Data used concerned production (dry matter intake, milk yield, body weight, body condition score, milk components) and reproduction (insemination and parturition time) data as well as rations’ composition. Preliminary results indicated that the adjustment with the new model was more accurate for the three production variables studied (dry matter intake, milk yield and body weight) during the first lactation than the one with the model without ‘feed plan’ module. Based on the first lactation fitting, the new model was also able to fit the second lactation data to a certain extend. To conclude, this first approach indicated the relevance of the feed module to improve the model fitting. It could therefore serve as management tool and help predicting the individual productivity of the next lactation. Further work should include perturbations (i.e. health incidents) in the model to obtain more accurate predictions.