The objective of this study was to determine the impact of a range of management systems on weed communities in cropped fields on farms in Saskatchewan. Farms (n = 28) with management systems defined by a combination of different cropping histories and chemical input levels were selected. Fields were surveyed on each farm in 1995, 1996, and 1997 after any post-emergent weed control to ensure that their weed communities reflected the impact of all agronomic management practices typical of the management system. Canonical correspondence analysis was used to determine whether weed communities were significantly correlated with management system after removal of variance due to ecoregion and year. The largest difference in weed communities attributable to farming system was between the systems with annual cropping histories and those that included perennials in the cropping history. Thus, the life history of the weeds reflected the life history of the crops. Herbicide use was correlated with the next largest difference between systems. A significant association between weed communities and different management systems indicates that weed species are being selected for by these systems. If such selection pressure continues, these species may become a threat to the system's sustainability. Key words: Weed communities, chemical input level, cropping history, canonical correspondence analysis, correspondence analysis, farming system