•The physical landslide model LAPSUS-LS was suitable for modelling the influence of vegetation on slope stability at the catchment scale.•The depth of the potential shear surface and root density were the most important parameters determining the volume of soil displaced during a landslide.•When the shear surface was 1.5 m deep, mixed coffee and tree plantations improved stability compared to coffee because of higher deep root density•When the shear surface was 1.0 m deep, no differences in stability were found between plantations as shallow rooting densities were similar. The influence of vegetation on slope stability is well understood at the slope level but scaling up to the catchment level is still a challenge, partially because of a lack of suitable data to validate models. We tested the physical landslide model, LAPSUS_LS, which models slope stability at the catchment scale. LAPSUS_LS combines a hydrological model with a Limit Equilibrium Method model, and calculates the factor of safety of individual cells based on their hydrological and geomorphological characteristics. We tested two types of vegetation on slope stability: (i) coffee monoculture (Coffea arabica) and (ii) a mixed plantation of coffee and deep rooting Erythrina (Erythrina poeppigiana) trees. Using soil and root data from Costa Rica, we performed simulations to test the response of LAPSUS_LS to root reinforcement, soil bulk density, transmissivity, internal friction angle and depth of shear plane. Furthermore, we modified the model to include biomass surcharge effect in the calculations. Results show that LAPSUS_LS was most sensitive to changes in additional cohesion from roots. When the depth of the shear plane was fixed at 1.0m, slopes were not unstable. However, when the shear plane was fixed to 1.5m, the mixed plantation of coffee and trees stabilized slopes, but the coffee monoculture was highly unstable, because root reinforcement was low at a depth of 1.5m. Soil transmissivity had a limited impact on the results compared to bulk density and internal friction angle. Biomass surcharge did not have any significant effect on the simulations. In conclusion, LAPSUS_LS responded well to the soil and vegetation input data, and is a suitable candidate for modeling the stability of vegetated slopes at the catchment level.