Black leaf streak disease (BLSD) caused by Mycosphaerella fijiensis is the most destructive leaf disease of bananas and plantains. Genetic improvement for resistance appears as the most appropriate control. As a high level of diversity is maintained in pathogen populations, partial resistance supposed durable is used instead of total resistance in breeding programmes. The aim of our studies is to evaluate the efficiency and durability of partial resistance to assist breeding programmes. To achieve this objective, three complementary approaches were undertaken: (i) Characterisation of partial resistance was realised by measuring the fungus life cycle parameters under both field and controlled conditions. The evaluation of 13 partial resistant varieties showed the existence of several components of resistance acting on different stages of the infectious cycle. (ii) The efficiency of two resistant varieties which differ for two resistance components (infection efficacy, ascospores production) were studied using epidemiological parameters. No difference in disease spatial dispersion and incidence was observed between resistant varieties during the first year whereas low differences in disease severity were measured. Moreover, these differences increased within years. These results could be explained by differences in production endogeneous inoculum. Experiments are conduced to measure endogeneous inoculum in each field to confirm this hypothesis. To test effects of other resistant components on BLS epidemics, mathematical modelling is being developed. The model is firstly based on disease description at field scale. (iii) Resistance durability is being studied through analysis of pathogen populations evolution. Molecular characterisation using CAPS markers was used on populations isolated after 6 and 25 months of cultivation. No significant difference between two populations coming from susceptible and resistant bananas was observed after 6 months. Pathogenicity variabi