Black Sigatoka (BS), caused by the fungal pathogen Mycosphaerella fijiensis, is considered as the most destructive foliar disease of bananas and plantains. Due to important damages on yield, the BS integrated management particularly by using resistant varieties appears essential. To advance our knowledge of pathogen dynamics and to design control method at plant scale, a mechanistic model was designed, calibrated and evaluated. The SiBaToKa model simulates under optimal epidemiological conditions the BS development on a banana by describing the disease severity within time at leaf and plant scale. It is developed in discrete time and it describes the BS lesions development during several crop cycles. Two sub-models were defined: the first one describes simply the banana growth in a deterministic way; the second one describes the complete and detailed epidemic cycle (including infection, lesions growth, asexual and sexual sporulations, spores dispersal). Numerical experiments and model performed sensitivity analyses enabled to identify the main influence of three epidemic parameters (lesions extension speed, incubation period, infection efficiency) and two parameters related to banana growth (the leaves number per plant, the leaf emission rate). The model was calibrated on plantain with available data and data of lesions kinetics on leaves acquired in Dominican Republic. The evaluation of the model, using an independent data set, highlighted the good quality of model predictions. Simulations allowed to evaluate the impact of host resistance components, the impact of auto-infection at plant scale, and the impact of the leaf emission rate on BS severity. The model allowed to better understand BS epidemic and it will allow to design virtually some efficient control practices at banana scale (such as deleafing, varietal resistance).