In the Mid-West region of Madagascar, yields in upland rice remain low due to both low soil fertility and high weed pressure in fields. Increasing managed biodiversity inside crop rotation may be an option to reduce weed pressure in cropping systems. The effect of managed biodiversity was studied in a field experiment carried out in Ivory (19°33'18.90"S, 46°24'53.08"E), with a randomized block design with four replications during two years. Three rainfed rice based rotations (Rice // Groundnut = RA, Rice // Sorghum + Vigna unguiculata =RSV, Rice//Mucuna cochinchinensis + Crotalaria spectabilis = RMC) combined with two levels of fertilization (Low Fertilization = manure vs. High Fertilization = manure + fertilizer) were compared to a rainfed rice monoculture. Each crop or crop mixture in the rotation was grown every year on plots measuring 45.9 m². Each year, weed and rice biomass were measured at each weeding date. Rice biomass was also measured at flowering and harvest. Rice yield and its components were measured at harvest. Weed flora was observed the second year to analyse potential change in weed communities according to fertilization and/or crop rotation. The first year, weed biomass and rice yield were higher in highly fertilized treatments than in lowly ones. In highly fertilized treatments, weeds did not affect yield and weed biomass was positively correlated to rice yield. On contrary, fertilization had no effect on weed biomass the second year. The highest and lowest weed biomass were observed on RA and RMC treatments respectively. Lowest rice yields were observed in RSV and RA treatments, in highly and lowly fertilized treatments respectively. In lowly fertilized treatments, weed biomass reduces significantly rice yield. Finally, rotation had more impact than fertilization on weed flora. Biodiversity may have positive effect on cropping systems performances but it depends largely on the introduced species.