Individual tree growth models need to integrate more processes, particularly carbon (C) and nitrogen (N) interactions, to gain more versatility and thus to better evaluate the responses of trees to global changes of climate and management practices. It is known that C source (leaf photosynthetic assimilation rate, Anet) and sinks (dark respiration, primary and secondary growth) are dependent to N content, but little is known on C reserves (i.e. starch) responses to N fertilization on apple trees. The present study focused on the C source-sink relationships in young apple trees (Malus pumila Mill., cv. Delbard Jubilé®), especially seasonal dynamics of both Anet and the carbohydrate reserves, in response to N supply. The responses to N fertilization were investigated through a combined experimental and modelling approach. Three levels of fertilization (none, normal, double) were applied in 2013 and 2014, each level being applied on three apple trees. During the same period, all trees were characterized in leaf gas exchanges and leaf N content. Current-year shoot (CYS) growth rate, in shoot length, number of leaf and leaf surface area was also characterized. At various dates, CYS were sampled for determining their carbohydrate content. Two apple trees (one “none” and one “double” treatment) were 3D-digitized in order to acquire the complete initial and final architecture during the growing season in 2013, as well as the aboveground biomass. Meteorological data (radiation, temperature, humidity, wind speed) were acquired by a weather station on the experimental site. The leaf gas exchange and CYS geometry and demography data were used to parameterize the RATP model (Sinoquet et al., 2001) for assessing the impact of the N fertilization on the source strength. The experimental results showed a significant impact of the N fertilization on the Farquhar sub-model parameters, but not on the Jarvis sub-model ones of the RATP model. Among the architectural variables, the CYS length decreased in the “none” treatment. The seasonal dynamics of the carbohydrate reserves was significantly influenced by the N treatment: the minimal value of starch content was reached in late spring, with a significantly higher level for the “none” treatment than for the other treatments, even during the summer. The RATP model is integrated within the C-based model PIAF-1 (Lacointe and Donès, 2007). Thus, the growth rate and carbohydrate data will be used to parameterize the PIAF-1 model to simulate local growth and reserve dynamics. PIAF-1 will enable to determine at which extend the lack of N limited the source activity and the sink strength.