Background: Invasions of new races can have contrasted consequences on populations of Puccinia striiformis f.sp. tritici (Pst, causing stripe rust of wheat). For example, the emergence of the Warrior race (PstS7), which was detected in France in 2011, had major impacts in Europe and in France. By contrast, PstS2 , which was detected in France in 2004, but had no impact in this country, contrary to other parts of the world. The objective of this study was to better understand factors that govern the success of an invasive race, considering these two races as case studies. We compared these races based on three factors: (i) virulence against local cultivars, (ii) aggressiveness in local environmental conditions and (iii) competitiveness against local races. Research findings: PstS2 has a relatively narrow virulence spectrum. Cultivars with at least one known resistance gene providing protection against PstS2 represented 70% of the grown area over the period during which PstS2 was detected (2003-2007). PstS7 has a much wider virulence spectrum but it was difficult to determine whether cultivars could be attacked by this race. We focused on the 48 most cultivated cultivars representing 77% of bread wheat area over the considered period (2009-2013). We gathered different types of experimental results, including varietal trials in natural or artificial inoculation, list of cultivars on which PstS7 isolates were collected (as part of the French national Pst population surveys) and seedling tests with PstS7. Based on this information, we were able to classify about 50% of studied cultivars as having a low or high risk of attack by PstS7. These cultivars represented respectively 17% (high risk) and 15% (low risk) of the wheat area. Isolates from different types (local: North PstS0 and South isolates PstS3 ; invasive: PstS7 and PstS2 isolates) were compared for infection efficiency and latency period in different temperature conditions. Thermal response curves were adjusted for both aggressiveness traits. Significant differences were found between isolate types in terms of latency period but not of infection efficiency. PstS7 and PstS2 were found to combine advantages compared to local isolates: on one hand a warmer optimal temperature and a shorter latency period at optimal temperature than North PstS0 isolates and, on the other hand, a wider thermal performance breadth than South isolates. Competition was assessed in a controlled condition experiment at seedling stage and at optimal temperature. Four combinations of isolates inoculated in equiproportion were tested with two repetitions each. Four invasive isolates (2 PstS7 and 2 PstS2) and 2 local North isolates were included. Final proportion of the invasive isolates was assessed after 6 cycles. This proportion was high for PstS7 isolates (between 61 and 84%) and variable for PstS2 isolates (between 18 and 82%). In silico experiments were carried out based on thermal response curves established for infection efficiency. Results were consistent with experimental data. PstS7 was more competitive over a large range of temperature conditions, while PstS2 was competitive only close to the optimal temperature. Finally, we assessed the potential number of cycles for different types of isolates, based on their individual thermal response curves for latency period and daily temperature recorded in 20 French sites x 10 years. In all sites, PstS2 isolates were always the fastest. The ranking of PstS7 isolates relative to local isolates (2 North and 2 South) was variable, but there was always at least one PstS7 isolate (among seven that were tested) that performed better than all local isolates. Conclusions: Our findings were consistent with previous hypotheses (eg. adaptation to high temperature, virulence spectrum) formulated to explain observed emergence events and we emphasised key features that emerge when considering a global analysis. Though estimating varietal risk was not trivial, combining different sources of information provided valuable information. Aggressiveness, in particular latency period, was useful to estimate the competitive ability of the considered races. We observed that adaptation at the population level could play an important role in overall competitive ability. This work made it possible to better understand factors involved in the emergence of new races of Pst, that can cause dramatic epidemics of stripe rust.