Weed data were collected over three tillage systems and two years in southwestern Ontario, in fields of corn, soybean, and winter wheat. Tillage systems involved soil inversion by moldboard plow, soil disturbance without inversion, and no tillage. The data were subjected to multivariate ordination techniques to determine if weed communities were associated with the different crops, tillage systems, or years. Crop, year, and tillage system all had significant effects on the weed community. Weed communities in corn and winter wheat were least similar; those in soybean were intermediate. Several grass weeds were more abundant in the drier of the two years. Some variation was related to tillage system but the effect of tillage was less than that of either crop or year.