Conventionally, coffee farmers try to control American leaf spot disease by applying treatments directed to residual inoculum present on old damaged coffee leaves. However, the effectiveness of these treatments is low, particularly in Niña years where the epidemics increase faster. Our hypothesis is that there is a "hidden inoculum" not controlled, living all the year, in the companion vegetation (weeds and shade trees) which can efficiently contribute to the epidemic onset. In this study, we assessed the aggressiveness of different strains we isolated from several hosts found in coffee plantations. We collected leaves with symptoms or/and gemmae from shade trees, weeds, Caturra (McCa) and Catimor (McK) coffee varieties. These samples were transported to the University of Costa Rica Phytopathology Lab for pathogen isolation or for direct inoculum picking. At least 15 gemmae, one by inoculation point, were placed on the surface of healthy and fresh Caturra leaves of two years old. Then, these were maintained into humid boxes at 20-21°C and 100% of relative humidity for 15 days, and at 24-25°C and approximatively 80% of relative humidity for other two weeks. During this period, the number of formed lesions, the mean diameter of each lesion and the quantity of geminifers produced were assessed each two days. Infection success (IS), gemmae production capacity, incubation and latent periods, and the pathogenicity index (PI) were calculated for each strain. Inoculum directly coming from field was more aggressive than inoculum coming from in vitro culture. The strain McK had a 100% IS; all its gemmae were viable; it was the first in producing lesions (2 days before the others); it also started the geminifers production 6 days before McCa. The isolates with major PI were recovered from Anredera cordifolia (9.1), Catimor (9.9) and Bryophyllum calycinum (18.9). Caturra only showed a PI of 3.67, reflecting a low pathogenicity. The main conclusions were: 1) There are difference