Clements, D. R. and DiTommaso, A. 2012. Predicting weed invasion in Canada under climate change: Evaluating evolutionary potential. Can. J. Plant Sci. 92: 1013-1020. Many weed species have already advanced northward from the United States into Canada, and their number threatens to increase with warming trends under climate change. For many weed species, this range expansion can be attributed to evolutionary adaptation by northern populations occurring in areas experiencing warmer climates and longer growing seasons in recent decades. In this paper, we examine the potential for invasive plant species to be selected for one or more of 10 character traits: (1) high growth rate, (2) wide climatic or environmental tolerance, (3) short generation time, (4) prolific or consistent reproduction, (5) small seed size, (6) effective dispersal, (7) uniparental reproduction capacity, (8) no specialized germination requirements, (9) high competitive ability, and (10) effective defenses against natural enemies. If any one of these traits is selected for in an invasive species, it would provide the affected species with the potential for a more expansive invasion range than anticipated by models that assume a static genotype. Four weed species of interest exhibiting potential northward range expansion within North America were evaluated: an obligatory outcrossing annual dicot, Himalayan balsam (Impatiens glandilifera Royle), a mostly selfing annual dicot, velvetleaf (Abutilon theophrasti Medic.), a perennial dicot that reproduces via rhizomes but forms fertile hybrids, Japanese knotweed [Fallopia japonica (Houtt.) Ronse Decr.], and a primarily selfing perennial grass, johnsongrass [Sorghum halapense (L.) Pers.]. Evidence for potential evolutionary responses to climate change was observed among particular traits for each of the four species, despite population genetic differences. The population genetics of invasive plants are difficult to model, as is climate change itself, but consideration of weed evolution to whatever degree possible should lead to improved predictive power of such models.