Intercropping, i.e., growing several species in the same field for a major part of their growing periods, often improves yield and weed control, but their performance greatly varies across situations. The aim of this study was to evaluate the effects of bi-species legume–cereal intercrops on weed dynamics and their impact on crop production, in the absence of nitrogen or water stress, via simulations with FLORSYS. This individual-based 3D model simulates daily crop–weed seed and plant dynamics over the years, from cropping system and pedoclimate, focusing on competition for light. The study tested seven species proportions in two species mixtures (wheat–faba bean and barley–pea) and nine spatial sowing patterns in three species mixtures (triticale–faba bean, wheat–faba bean, wheat–pea), in both cases comparing the intercrops with the corresponding sole crops (controls). Intercrops and controls were inserted into rotations and simulated over 30 years and repeated with 10 climate scenarios from South-Western France, either with or without weeds. The simulations showed that: (1) the intercrops that best controlled weeds were barley–pea and triticale–faba bean, (2) the spatial pattern alternating one cereal row with one legume row as well as the 67 %-cereal–33 %-legume and 100 %-cereal–50 %-legume species proportions were those that maximised yields and minimised losses due to weeds, (3) the weed biomass in intercrop was greater than or equal to that of the sole cereal, and less than that of the sole legume, and (4) legumes benefitted more from intercropping than cereals because cereals are more competitive against weeds. Intercrop yield was best when combining species with contrasting shading responses (etiolated with stockier plants, leafy with stemmier plants) but early and good plant emergence was essential, particularly for weed suppression.