This work investigates the dynamic operation of a large solid oxide fuel cell (SOFC) system using the system identification modeling approach. Transient modeling can be of particular interest in order to describe and optimize start-up and shut-down procedures in SOFC systems; moreover, mathematical description of transients is helpful in the diagnostic of hazardous conditions during dynamic operation (i.e., load following). Physical based models—described by differential equations—are usually complex and need significant computing time. To achieve real-time capability, the focus is on empirical models. In this study, the transient operation of a large SOFC generator is reproduced by using the system identification approach that is based on the definition of a black-box model and on the identification of main model coefficients based on real experimental data. In particular, the start-up of the system and its dynamics under fuel sensitivity experiments are investigated. First, the work valuably shows some experimental results of a large SOFC generator under dynamic operation. In particular, modeling results show that the system identification approach is an effective tool for the description of the transient behavior of SOFC large systems. The simulation of the start-up of a 100 kWe (electric) SOFC system shows the possibility to save ∼20% of start-up time using an electric air heater rated a 30% in power than the nominal one. The increase of natural gas flow during the start-up operation is instead not beneficial in term of yielding a faster procedure; rather, it leads to an excessive cooling of the stack due to internal reforming of natural gas (the higher amount of fuel burnt in the combustion zone does not lead to a more efficient way to preheat the cathode air). The results of this work are fully supported by experimental data available from a real generator running in Italy through years 2006–2009 and thus believed to be a valuable contribution for the scientific and engineering community.